1
00:00:11,599 --> 00:00:17,920
hello and welcome back

2
00:00:14,400 --> 00:00:17,920
to pike online

3
00:00:19,359 --> 00:00:23,920
it's been so good

4
00:00:20,880 --> 00:00:25,840
fantastic it's like yeah

5
00:00:23,920 --> 00:00:27,680
i've had an absolute blast

6
00:00:25,840 --> 00:00:31,119
i've lost for words i'm lost for words

7
00:00:27,680 --> 00:00:32,800
i've lost my words they're all gone um

8
00:00:31,119 --> 00:00:35,520
but that's okay because i don't have to

9
00:00:32,800 --> 00:00:36,559
talk right now we get to introduce

10
00:00:35,520 --> 00:00:38,399
um

11
00:00:36,559 --> 00:00:40,719
the i can't say it's the coolest part of

12
00:00:38,399 --> 00:00:43,280
the day because it's just one of the

13
00:00:40,719 --> 00:00:45,039
coolest parts of the coolest day

14
00:00:43,280 --> 00:00:46,960
yes

15
00:00:45,039 --> 00:00:48,320
amanda tell me about the student

16
00:00:46,960 --> 00:00:51,039
showcase

17
00:00:48,320 --> 00:00:52,800
okay so a couple of years ago when pycon

18
00:00:51,039 --> 00:00:55,680
was in sydney

19
00:00:52,800 --> 00:00:58,559
we had a couple of very keen students

20
00:00:55,680 --> 00:01:00,800
sign up for the main track of

21
00:00:58,559 --> 00:01:02,079
pycon which is a really exciting thing

22
00:01:00,800 --> 00:01:04,080
and um

23
00:01:02,079 --> 00:01:05,760
we thought maybe

24
00:01:04,080 --> 00:01:07,439
we should find a way to encourage this

25
00:01:05,760 --> 00:01:08,400
to happen more often

26
00:01:07,439 --> 00:01:12,000
indeed

27
00:01:08,400 --> 00:01:13,680
and so we invented the education track

28
00:01:12,000 --> 00:01:15,920
and we put it out to teachers in our

29
00:01:13,680 --> 00:01:18,720
network and tried to get as many people

30
00:01:15,920 --> 00:01:20,240
to enter as possible and it's just grown

31
00:01:18,720 --> 00:01:22,240
from there and we had the most i think

32
00:01:20,240 --> 00:01:23,439
we had the most applicants ever this

33
00:01:22,240 --> 00:01:26,560
year

34
00:01:23,439 --> 00:01:29,520
which was fantastic and terrible it's

35
00:01:26,560 --> 00:01:33,040
fantastic and terrible that's right

36
00:01:29,520 --> 00:01:35,360
we have such an awesome lineup of kids

37
00:01:33,040 --> 00:01:38,000
uh who want to talk to us today

38
00:01:35,360 --> 00:01:41,759
and these students have made fantastic

39
00:01:38,000 --> 00:01:43,200
things using python with python um

40
00:01:41,759 --> 00:01:44,479
they're going to tell us about all of

41
00:01:43,200 --> 00:01:46,079
the things that went right all the

42
00:01:44,479 --> 00:01:49,119
things that went wrong

43
00:01:46,079 --> 00:01:50,159
um some of them will be with us in the

44
00:01:49,119 --> 00:01:51,759
chat

45
00:01:50,159 --> 00:01:53,759
uh

46
00:01:51,759 --> 00:01:54,560
i can never tell which side it will be

47
00:01:53,759 --> 00:01:57,840
on

48
00:01:54,560 --> 00:02:00,079
um so you can ask questions there

49
00:01:57,840 --> 00:02:02,399
it's one of these directions

50
00:02:00,079 --> 00:02:02,399
um

51
00:02:04,159 --> 00:02:08,000
they're just so good

52
00:02:05,840 --> 00:02:09,280
yeah the kids are so good the talks are

53
00:02:08,000 --> 00:02:10,560
so good the kids are all right i think

54
00:02:09,280 --> 00:02:12,000
that's what i'm trying to say amanda the

55
00:02:10,560 --> 00:02:14,000
kids are all right they must have some

56
00:02:12,000 --> 00:02:15,680
pretty good teachers or something

57
00:02:14,000 --> 00:02:16,959
um i don't know i'm not just saying that

58
00:02:15,680 --> 00:02:19,120
because i know some of them are in the

59
00:02:16,959 --> 00:02:21,120
audience now um and i'm not just saying

60
00:02:19,120 --> 00:02:23,520
that in order to to spruik

61
00:02:21,120 --> 00:02:25,520
next year when the student showcase will

62
00:02:23,520 --> 00:02:28,239
also happen teachers

63
00:02:25,520 --> 00:02:30,800
get thinking now um

64
00:02:28,239 --> 00:02:32,720
the rules are that you have to be a high

65
00:02:30,800 --> 00:02:34,840
school student and you have to have done

66
00:02:32,720 --> 00:02:37,840
a thing that involves

67
00:02:34,840 --> 00:02:40,480
python that's it that's pretty low bar

68
00:02:37,840 --> 00:02:43,599
we can fall over that bar in indeed

69
00:02:40,480 --> 00:02:45,840
indeed um so while you are all sitting

70
00:02:43,599 --> 00:02:48,720
at home reclining watching these amazing

71
00:02:45,840 --> 00:02:50,080
kids talk about their amazing projects

72
00:02:48,720 --> 00:02:52,239
think about

73
00:02:50,080 --> 00:02:54,160
what you dear viewer or

74
00:02:52,239 --> 00:02:56,640
student or teacher could do for next

75
00:02:54,160 --> 00:02:58,640
year to perhaps

76
00:02:56,640 --> 00:03:02,080
see yourselves or your students

77
00:02:58,640 --> 00:03:02,080
presenting right here

78
00:03:02,239 --> 00:03:06,480
without further ado i'm going to let

79
00:03:04,640 --> 00:03:09,200
these fantastic students introduce

80
00:03:06,480 --> 00:03:12,080
themselves by which i mean

81
00:03:09,200 --> 00:03:15,200
i'm going to throw to past nicki

82
00:03:12,080 --> 00:03:19,720
uh to to make my current life easier all

83
00:03:15,200 --> 00:03:19,720
right uh thanks past nikki

84
00:03:24,720 --> 00:03:28,640
thanks future nikki

85
00:03:26,400 --> 00:03:30,239
our first speaker today is liv and

86
00:03:28,640 --> 00:03:31,519
she'll be talking about renewable

87
00:03:30,239 --> 00:03:33,440
hearing

88
00:03:31,519 --> 00:03:35,040
liv has been programming for eight years

89
00:03:33,440 --> 00:03:37,280
and when i asked what her favorite emoji

90
00:03:35,040 --> 00:03:39,760
is she said it's the saturn emoji or the

91
00:03:37,280 --> 00:03:41,519
ringed planet emoji

92
00:03:39,760 --> 00:03:43,519
liv has recently been learning about

93
00:03:41,519 --> 00:03:45,680
machine learning and data analysis and

94
00:03:43,519 --> 00:03:46,400
i'm really excited to hear her talk

95
00:03:45,680 --> 00:03:48,480
hi

96
00:03:46,400 --> 00:03:50,480
my name is liv and i'm a student at

97
00:03:48,480 --> 00:03:53,280
lorry novala steiner school i have been

98
00:03:50,480 --> 00:03:54,799
programming python for eight years now

99
00:03:53,280 --> 00:03:56,000
it started when i broke my foot when i

100
00:03:54,799 --> 00:03:58,879
was 11

101
00:03:56,000 --> 00:04:01,200
and i couldn't walk for a few weeks

102
00:03:58,879 --> 00:04:03,599
so i started learning python from a book

103
00:04:01,200 --> 00:04:05,599
and started creating projects like guess

104
00:04:03,599 --> 00:04:06,799
the number and choose your own adventure

105
00:04:05,599 --> 00:04:08,799
style games

106
00:04:06,799 --> 00:04:10,640
at my school in our final year we

107
00:04:08,799 --> 00:04:12,959
undertake a major work

108
00:04:10,640 --> 00:04:14,879
this means we choose a topic to study in

109
00:04:12,959 --> 00:04:16,959
depth and then presented at the end of

110
00:04:14,879 --> 00:04:19,359
the year we complete a practical

111
00:04:16,959 --> 00:04:21,359
physical component and a written thesis

112
00:04:19,359 --> 00:04:24,680
of eight to ten thousand words and a

113
00:04:21,359 --> 00:04:24,680
final presentation

114
00:04:32,720 --> 00:04:36,960
i chose to study engineering because in

115
00:04:35,199 --> 00:04:40,080
a world where everything is increasingly

116
00:04:36,960 --> 00:04:42,639
digital i believe that technology holds

117
00:04:40,080 --> 00:04:44,960
the key to an equal and sustainable

118
00:04:42,639 --> 00:04:46,960
future the physical component of my

119
00:04:44,960 --> 00:04:50,000
major works is a solar-powered hearing

120
00:04:46,960 --> 00:04:52,240
aid in my own experience hearing aids go

121
00:04:50,000 --> 00:04:54,080
through batteries every two weeks having

122
00:04:52,240 --> 00:04:55,919
worn a hearing aid since i was about

123
00:04:54,080 --> 00:04:59,120
eight years old that means that i've

124
00:04:55,919 --> 00:05:02,800
already gone through 286 batteries and

125
00:04:59,120 --> 00:05:04,960
over 2500 over my entire life

126
00:05:02,800 --> 00:05:06,960
the impact of batteries on wildlife is

127
00:05:04,960 --> 00:05:08,720
incredibly detrimental

128
00:05:06,960 --> 00:05:11,360
battery sent to landfills will leak

129
00:05:08,720 --> 00:05:13,520
toxic substances into the ground which

130
00:05:11,360 --> 00:05:14,880
affects the nearby ecosystems and

131
00:05:13,520 --> 00:05:16,800
waterways

132
00:05:14,880 --> 00:05:18,240
carelessly discarding batteries will

133
00:05:16,800 --> 00:05:21,120
result in

134
00:05:18,240 --> 00:05:24,479
animals such as birds mistakingly eating

135
00:05:21,120 --> 00:05:26,320
the batteries and poisoning themselves

136
00:05:24,479 --> 00:05:28,800
the financial cost of batteries can get

137
00:05:26,320 --> 00:05:30,560
expensive replacing them every two weeks

138
00:05:28,800 --> 00:05:32,880
with batteries costing up to ten dollars

139
00:05:30,560 --> 00:05:35,360
a pack of eight batteries

140
00:05:32,880 --> 00:05:38,400
using a single rechargeable battery cuts

141
00:05:35,360 --> 00:05:40,080
this cost entirely my thesis focuses on

142
00:05:38,400 --> 00:05:42,240
how hearing loss affects indigenous

143
00:05:40,080 --> 00:05:44,400
australians as their rates of hearing

144
00:05:42,240 --> 00:05:46,880
loss are nearly three times that of

145
00:05:44,400 --> 00:05:49,120
non-indigenous people this statistic is

146
00:05:46,880 --> 00:05:51,199
particularly prevalent in children as

147
00:05:49,120 --> 00:05:53,199
they are more likely to contract otitis

148
00:05:51,199 --> 00:05:55,600
media an ear infection that can have

149
00:05:53,199 --> 00:05:57,440
lifelong after effects numerous studies

150
00:05:55,600 --> 00:06:00,479
have found that hearing loss at a young

151
00:05:57,440 --> 00:06:02,720
formative age impacts social mental and

152
00:06:00,479 --> 00:06:04,479
cognitive wellbeing the development of

153
00:06:02,720 --> 00:06:06,560
higher cognitive thinking and

154
00:06:04,479 --> 00:06:09,280
interpersonal relationships occurs from

155
00:06:06,560 --> 00:06:12,639
early childhood to adolescence and

156
00:06:09,280 --> 00:06:14,160
hearing loss can severely impact this

157
00:06:12,639 --> 00:06:16,639
the hearing aid is specifically built

158
00:06:14,160 --> 00:06:18,720
with indigenous populations in mind

159
00:06:16,639 --> 00:06:20,560
as remote communities may not have

160
00:06:18,720 --> 00:06:22,800
access to batteries or be able to afford

161
00:06:20,560 --> 00:06:24,960
them the original development of the

162
00:06:22,800 --> 00:06:27,280
hearing aid was on a raspberry pi picot

163
00:06:24,960 --> 00:06:30,639
board the board was very new being

164
00:06:27,280 --> 00:06:32,960
released in january 2021 but had very

165
00:06:30,639 --> 00:06:35,039
high processing speed i coded it in

166
00:06:32,960 --> 00:06:37,360
circuit python as i needed some more

167
00:06:35,039 --> 00:06:39,280
heavy duty libraries i started by

168
00:06:37,360 --> 00:06:41,759
calculating the decibels and then of

169
00:06:39,280 --> 00:06:43,440
course a basic blink script

170
00:06:41,759 --> 00:06:45,680
of course i would need to do a little

171
00:06:43,440 --> 00:06:47,440
more for a functioning hearing aid since

172
00:06:45,680 --> 00:06:49,199
not every deaf person has the same

173
00:06:47,440 --> 00:06:51,840
degree of hearing loss i would need to

174
00:06:49,199 --> 00:06:54,160
perform a fast fourier transform

175
00:06:51,840 --> 00:06:56,960
hearing loss is not uniform in severity

176
00:06:54,160 --> 00:06:58,639
or range some may be able to hear higher

177
00:06:56,960 --> 00:07:01,199
frequencies better than lower

178
00:06:58,639 --> 00:07:03,599
frequencies or vice versa to create the

179
00:07:01,199 --> 00:07:05,759
most beneficial hearing aid i would need

180
00:07:03,599 --> 00:07:08,319
to target the frequencies that the

181
00:07:05,759 --> 00:07:09,599
person couldn't hear

182
00:07:08,319 --> 00:07:11,680
to do this

183
00:07:09,599 --> 00:07:13,440
i needed to separate the frequencies in

184
00:07:11,680 --> 00:07:15,360
each audio sample

185
00:07:13,440 --> 00:07:18,000
a fourier transform essentially breaks

186
00:07:15,360 --> 00:07:20,000
down frequencies into their sine waves

187
00:07:18,000 --> 00:07:22,240
it helps to determine which frequencies

188
00:07:20,000 --> 00:07:25,440
in a sample i'm going to target needless

189
00:07:22,240 --> 00:07:26,479
to say it was very hard to quote one of

190
00:07:25,440 --> 00:07:28,080
my

191
00:07:26,479 --> 00:07:30,000
favorite books by josh barrows on

192
00:07:28,080 --> 00:07:32,720
fourier transforms

193
00:07:30,000 --> 00:07:34,639
don't ask really don't ask unless you're

194
00:07:32,720 --> 00:07:36,479
willing to put in a couple years of

195
00:07:34,639 --> 00:07:37,680
degree level maps i still have

196
00:07:36,479 --> 00:07:40,560
nightmares

197
00:07:37,680 --> 00:07:43,120
to be honest i still don't understand it

198
00:07:40,560 --> 00:07:44,960
after about two months i finally got it

199
00:07:43,120 --> 00:07:46,800
the microphone was picking up sound and

200
00:07:44,960 --> 00:07:49,199
calculating the transform of certain

201
00:07:46,800 --> 00:07:50,800
frequencies i couldn't believe it now

202
00:07:49,199 --> 00:07:53,599
all i had to do was send it to the

203
00:07:50,800 --> 00:07:56,879
output that's where i was wrong the pico

204
00:07:53,599 --> 00:07:59,840
even with a dual core arm cortex mo plus

205
00:07:56,879 --> 00:08:02,400
processor with 264 kilobytes internal

206
00:07:59,840 --> 00:08:03,440
ram and support for up to 60 megabytes

207
00:08:02,400 --> 00:08:06,000
of off

208
00:08:03,440 --> 00:08:09,199
chip flash it just didn't have the power

209
00:08:06,000 --> 00:08:11,599
to do proper audio processing

210
00:08:09,199 --> 00:08:13,599
i could do everything separately but as

211
00:08:11,599 --> 00:08:15,360
soon as i tried to put it all together

212
00:08:13,599 --> 00:08:17,360
it just didn't work

213
00:08:15,360 --> 00:08:19,599
i rang around in hopes that there might

214
00:08:17,360 --> 00:08:21,840
be something i could do but no

215
00:08:19,599 --> 00:08:24,720
everyone told me that i had to move on

216
00:08:21,840 --> 00:08:27,280
to something called the tinsy

217
00:08:24,720 --> 00:08:28,720
i was a little bit upset

218
00:08:27,280 --> 00:08:31,120
i had spent months trying to get it to

219
00:08:28,720 --> 00:08:33,039
work on python and figured out how to do

220
00:08:31,120 --> 00:08:34,719
a fourier transform only to find that it

221
00:08:33,039 --> 00:08:37,599
failed at the last step

222
00:08:34,719 --> 00:08:39,680
but there was no time to be sad so i

223
00:08:37,599 --> 00:08:41,839
bought a tendency and an audio shield

224
00:08:39,680 --> 00:08:44,880
and immediately blew them both up but

225
00:08:41,839 --> 00:08:49,810
there was no time to be sad so i bought

226
00:08:44,880 --> 00:09:24,669
another tinsy and this time it worked

227
00:08:49,810 --> 00:09:24,669
[Music]

228
00:09:27,260 --> 00:10:05,849
[Music]

229
00:10:08,890 --> 00:10:33,990
[Music]

230
00:10:35,600 --> 00:10:40,320
using the handy audio library built for

231
00:10:37,920 --> 00:10:42,640
tinsy and the graphic interface for

232
00:10:40,320 --> 00:10:44,720
writing code i managed to build

233
00:10:42,640 --> 00:10:47,440
something resembling a hearing aid the

234
00:10:44,720 --> 00:10:50,160
code for the fourier transform was so

235
00:10:47,440 --> 00:10:51,680
easy with the audio design system it was

236
00:10:50,160 --> 00:10:53,920
my saving grace

237
00:10:51,680 --> 00:10:55,920
i was able to send the audio via

238
00:10:53,920 --> 00:10:58,560
bluetooth to a receiver that sits in the

239
00:10:55,920 --> 00:11:01,360
air using an old hearing aid speaker i

240
00:10:58,560 --> 00:11:03,279
had lying around building a hearing aid

241
00:11:01,360 --> 00:11:04,959
from scratch wasn't easy but i never

242
00:11:03,279 --> 00:11:06,480
expected it to be

243
00:11:04,959 --> 00:11:09,120
growing up with hearing loss i

244
00:11:06,480 --> 00:11:11,600
understand how important it is to have

245
00:11:09,120 --> 00:11:13,279
the resources available so that i have

246
00:11:11,600 --> 00:11:15,279
the same chance as everyone else around

247
00:11:13,279 --> 00:11:17,600
me to go to school partake in

248
00:11:15,279 --> 00:11:20,240
discussions about my health and develop

249
00:11:17,600 --> 00:11:24,810
a sustainable and equal future and that

250
00:11:20,240 --> 00:11:29,120
is why i built renewable hearing

251
00:11:24,810 --> 00:11:30,880
[Music]

252
00:11:29,120 --> 00:11:32,240
okay thanks

253
00:11:30,880 --> 00:11:34,079
bye

254
00:11:32,240 --> 00:11:36,720
thank you so much liv

255
00:11:34,079 --> 00:11:38,959
up next is jonty jonty's talking about

256
00:11:36,720 --> 00:11:40,320
python meta programming through bytecode

257
00:11:38,959 --> 00:11:41,680
and has been learning to program for

258
00:11:40,320 --> 00:11:43,680
three years now

259
00:11:41,680 --> 00:11:45,920
his favorite emoji is the smiley face

260
00:11:43,680 --> 00:11:48,320
emoji although technically he sent

261
00:11:45,920 --> 00:11:50,800
through a colon in a bracket but i think

262
00:11:48,320 --> 00:11:52,160
i'm being a little pedantic here

263
00:11:50,800 --> 00:11:54,000
when i asked what he's been recently

264
00:11:52,160 --> 00:11:55,839
learning about computers he said he's

265
00:11:54,000 --> 00:11:57,360
been learning about the axioms of

266
00:11:55,839 --> 00:12:00,240
natural deduction

267
00:11:57,360 --> 00:12:03,279
let's hear his talk i am john t deakin

268
00:12:00,240 --> 00:12:06,000
and this is my project by code macro

269
00:12:03,279 --> 00:12:07,600
it's a look at intermediate languages in

270
00:12:06,000 --> 00:12:09,279
meta programming

271
00:12:07,600 --> 00:12:12,480
specifically in the context of like

272
00:12:09,279 --> 00:12:14,320
fourth and lisp macros um

273
00:12:12,480 --> 00:12:15,680
yeah so this is just a cool example i

274
00:12:14,320 --> 00:12:17,200
thought i'd flash up at the start we'll

275
00:12:15,680 --> 00:12:19,519
go through it um

276
00:12:17,200 --> 00:12:20,959
so i'm gonna go through some of the

277
00:12:19,519 --> 00:12:22,240
you know intermediate languages and meta

278
00:12:20,959 --> 00:12:24,320
programming stuff just make sure we're

279
00:12:22,240 --> 00:12:25,680
all on the same page before i start

280
00:12:24,320 --> 00:12:27,440
talking about

281
00:12:25,680 --> 00:12:29,440
the bytecode macro

282
00:12:27,440 --> 00:12:33,200
library itself um

283
00:12:29,440 --> 00:12:34,000
so intermediate languages what are they

284
00:12:33,200 --> 00:12:36,000
when

285
00:12:34,000 --> 00:12:37,920
a compiler runs it doesn't just run the

286
00:12:36,000 --> 00:12:39,440
string directly you know that would be

287
00:12:37,920 --> 00:12:41,360
insane it goes through a few steps

288
00:12:39,440 --> 00:12:43,040
beforehand

289
00:12:41,360 --> 00:12:45,200
and you know in this example here we've

290
00:12:43,040 --> 00:12:48,000
got is a string going through alexa to

291
00:12:45,200 --> 00:12:50,079
be uh to become tokens which is then

292
00:12:48,000 --> 00:12:52,240
passed to become an ast which is then

293
00:12:50,079 --> 00:12:53,839
compiled to become bytecode and then the

294
00:12:52,240 --> 00:12:56,079
by code's interpreted

295
00:12:53,839 --> 00:12:57,760
later um and it's that by code that's

296
00:12:56,079 --> 00:12:59,440
what we're looking at today

297
00:12:57,760 --> 00:13:01,440
um well there's not always by code this

298
00:12:59,440 --> 00:13:03,200
is just a overview of python itself it

299
00:13:01,440 --> 00:13:05,519
does you know changes from compiler to

300
00:13:03,200 --> 00:13:07,200
compiler

301
00:13:05,519 --> 00:13:10,240
but yeah so in python it's a byte code

302
00:13:07,200 --> 00:13:13,279
and a bytecode is like a series of bytes

303
00:13:10,240 --> 00:13:14,800
that is a shortened representation of

304
00:13:13,279 --> 00:13:16,720
the code and simplified and shortened

305
00:13:14,800 --> 00:13:18,079
representation of the code you can do

306
00:13:16,720 --> 00:13:20,160
some cool things but it's way easier to

307
00:13:18,079 --> 00:13:21,760
interpret and you can do some cool

308
00:13:20,160 --> 00:13:23,519
optimizations uh

309
00:13:21,760 --> 00:13:25,839
not so much with stack based by code

310
00:13:23,519 --> 00:13:28,160
which python is um and so a stack based

311
00:13:25,839 --> 00:13:29,360
back code is a byte code that primarily

312
00:13:28,160 --> 00:13:31,120
acts upon

313
00:13:29,360 --> 00:13:33,040
a stack

314
00:13:31,120 --> 00:13:34,720
a data set you know just all the other

315
00:13:33,040 --> 00:13:36,480
kinds of sex but mostly data's right the

316
00:13:34,720 --> 00:13:39,040
data sets what we're talking about um

317
00:13:36,480 --> 00:13:40,399
and so in this case we've got it loading

318
00:13:39,040 --> 00:13:42,800
uh on the right side here we've got it

319
00:13:40,399 --> 00:13:45,040
loading a string hello world onto the

320
00:13:42,800 --> 00:13:46,880
stack uh and then you can pop that

321
00:13:45,040 --> 00:13:48,800
string into send it out

322
00:13:46,880 --> 00:13:51,360
using the print command right

323
00:13:48,800 --> 00:13:52,880
um and so that's you know how python by

324
00:13:51,360 --> 00:13:54,639
code mostly works just a lot of these

325
00:13:52,880 --> 00:13:55,760
cool stack operations

326
00:13:54,639 --> 00:13:56,959
um

327
00:13:55,760 --> 00:13:57,839
and

328
00:13:56,959 --> 00:13:59,519
uh

329
00:13:57,839 --> 00:14:00,880
yeah and so we'll get into meta

330
00:13:59,519 --> 00:14:03,519
programming now

331
00:14:00,880 --> 00:14:05,199
meta programming is you know code acting

332
00:14:03,519 --> 00:14:06,720
upon code it's actually everywhere i i

333
00:14:05,199 --> 00:14:08,720
didn't realize this before it's checked

334
00:14:06,720 --> 00:14:10,079
but it's like almost every language is

335
00:14:08,720 --> 00:14:12,480
made up for some meta programming

336
00:14:10,079 --> 00:14:14,160
capabilities uh you know java c-sharp

337
00:14:12,480 --> 00:14:16,720
has reflection

338
00:14:14,160 --> 00:14:19,600
julie uh python has meta classes c plus

339
00:14:16,720 --> 00:14:21,360
plus famously template meta programming

340
00:14:19,600 --> 00:14:24,320
uh and fourth

341
00:14:21,360 --> 00:14:26,880
fourth uh lisp and julia

342
00:14:24,320 --> 00:14:29,360
all have this really cool kind of macros

343
00:14:26,880 --> 00:14:31,440
thing i mean fourth and julia are based

344
00:14:29,360 --> 00:14:33,199
you know inspired by this but

345
00:14:31,440 --> 00:14:34,320
they all have a really cool macro system

346
00:14:33,199 --> 00:14:35,279
where you can

347
00:14:34,320 --> 00:14:37,360
take

348
00:14:35,279 --> 00:14:40,000
uh any code you can take some code and

349
00:14:37,360 --> 00:14:41,839
then transform it into new code

350
00:14:40,000 --> 00:14:44,079
uh and

351
00:14:41,839 --> 00:14:46,240
you know you can extend the language you

352
00:14:44,079 --> 00:14:48,000
can write shorthands for some stuff you

353
00:14:46,240 --> 00:14:49,920
know with code reuse uh you can even

354
00:14:48,000 --> 00:14:52,320
like take it so far as to write domain

355
00:14:49,920 --> 00:14:54,399
specific languages in other language in

356
00:14:52,320 --> 00:14:56,560
your own language i think it's very cool

357
00:14:54,399 --> 00:14:59,519
uh so for example like in fixed maths

358
00:14:56,560 --> 00:15:01,760
you can do infinix maths in lisp

359
00:14:59,519 --> 00:15:03,600
some guys done that i think it's cool

360
00:15:01,760 --> 00:15:05,600
and for this project i specifically was

361
00:15:03,600 --> 00:15:06,800
looking mostly at fourth macros just

362
00:15:05,600 --> 00:15:08,560
because i think

363
00:15:06,800 --> 00:15:11,040
they've applied better and you know

364
00:15:08,560 --> 00:15:13,040
that's what i was learning at the time

365
00:15:11,040 --> 00:15:15,440
and the question is why if these are so

366
00:15:13,040 --> 00:15:16,720
cool and such as you know they're very

367
00:15:15,440 --> 00:15:19,120
beloved by

368
00:15:16,720 --> 00:15:21,199
list programmers especially macros very

369
00:15:19,120 --> 00:15:23,519
much loved by list programmers um

370
00:15:21,199 --> 00:15:25,519
and if they're so great why aren't they

371
00:15:23,519 --> 00:15:27,519
in other languages as much

372
00:15:25,519 --> 00:15:29,920
and i think in my opinion at least it's

373
00:15:27,519 --> 00:15:32,000
mostly because lisp has such a beautiful

374
00:15:29,920 --> 00:15:34,000
like regular syntax

375
00:15:32,000 --> 00:15:36,079
uh that it makes it a really easy

376
00:15:34,000 --> 00:15:37,600
interface to be manipulating through the

377
00:15:36,079 --> 00:15:39,839
meta programming right

378
00:15:37,600 --> 00:15:41,759
you know it's all lists embed lists and

379
00:15:39,839 --> 00:15:43,839
lists you know just recursive lists it's

380
00:15:41,759 --> 00:15:46,079
very easy to iterate through that kind

381
00:15:43,839 --> 00:15:48,000
of stuff same thing with fourth

382
00:15:46,079 --> 00:15:49,680
and that's why

383
00:15:48,000 --> 00:15:51,360
it works so well and i think that's

384
00:15:49,680 --> 00:15:52,399
where a lot of the libraries that try to

385
00:15:51,360 --> 00:15:54,399
implement this kind of stuff in

386
00:15:52,399 --> 00:15:55,759
languages that weren't designed for it

387
00:15:54,399 --> 00:15:57,680
like fall down

388
00:15:55,759 --> 00:15:59,360
because they're trying to use the ast

389
00:15:57,680 --> 00:16:02,000
representation that lisp uses in other

390
00:15:59,360 --> 00:16:03,279
languages with a really complex ast

391
00:16:02,000 --> 00:16:04,880
and it makes it you know just makes it

392
00:16:03,279 --> 00:16:06,880
confusing and hard to

393
00:16:04,880 --> 00:16:08,079
reason about in my opinion at least um

394
00:16:06,880 --> 00:16:10,959
and and so that's what i was trying to

395
00:16:08,079 --> 00:16:13,519
do with this project was to capture that

396
00:16:10,959 --> 00:16:15,360
simplicity in the interface that lisp

397
00:16:13,519 --> 00:16:17,279
has that makes it so good

398
00:16:15,360 --> 00:16:20,880
uh and so yeah speaking of we're gonna

399
00:16:17,279 --> 00:16:20,880
get into bytecode macro

400
00:16:20,959 --> 00:16:28,720
so how bytecode macro works is it takes

401
00:16:25,759 --> 00:16:31,279
you know it's a decorator that takes a

402
00:16:28,720 --> 00:16:33,440
macro function and the function to apply

403
00:16:31,279 --> 00:16:35,199
that macro to

404
00:16:33,440 --> 00:16:37,680
converts that

405
00:16:35,199 --> 00:16:41,519
function it passed in this case test to

406
00:16:37,680 --> 00:16:43,519
a list of tuples or tuples

407
00:16:41,519 --> 00:16:45,920
uh and

408
00:16:43,519 --> 00:16:46,880
you know those tuples are instructions

409
00:16:45,920 --> 00:16:48,959
in the bike

410
00:16:46,880 --> 00:16:51,279
in the in the in the buy code right it's

411
00:16:48,959 --> 00:16:53,759
a simplified human readable version

412
00:16:51,279 --> 00:16:54,959
of that by code and it passes this list

413
00:16:53,759 --> 00:16:56,320
of tuples

414
00:16:54,959 --> 00:16:59,680
to the

415
00:16:56,320 --> 00:17:02,480
macro function in this case macro trace

416
00:16:59,680 --> 00:17:04,400
and then that macro function can

417
00:17:02,480 --> 00:17:06,400
modify that by code

418
00:17:04,400 --> 00:17:08,079
that linear list of tuples to be

419
00:17:06,400 --> 00:17:10,000
whatever it wanted

420
00:17:08,079 --> 00:17:12,640
so in this case we've changed hello

421
00:17:10,000 --> 00:17:15,520
world to hellofrance

422
00:17:12,640 --> 00:17:16,959
and then you return this newly modified

423
00:17:15,520 --> 00:17:19,199
list of tuples

424
00:17:16,959 --> 00:17:20,640
and it will execute this as if it was

425
00:17:19,199 --> 00:17:23,039
the original function

426
00:17:20,640 --> 00:17:24,640
um and there we go this is saying hello

427
00:17:23,039 --> 00:17:26,480
friends at the end

428
00:17:24,640 --> 00:17:27,839
um and this

429
00:17:26,480 --> 00:17:29,520
i think that's really cool and that's

430
00:17:27,839 --> 00:17:30,799
that's how it works is very simple i

431
00:17:29,520 --> 00:17:32,880
think it's you know easy enough to

432
00:17:30,799 --> 00:17:34,400
understand and this is the

433
00:17:32,880 --> 00:17:36,880
macro that was used to produce the

434
00:17:34,400 --> 00:17:39,919
visualizations we saw just

435
00:17:36,880 --> 00:17:42,160
last slide uh it takes in the tuples and

436
00:17:39,919 --> 00:17:45,280
then 4n in the tuples of prince enzo

437
00:17:42,160 --> 00:17:46,720
prints each tuple in a line

438
00:17:45,280 --> 00:17:47,440
and then it returns it without changing

439
00:17:46,720 --> 00:17:49,840
it

440
00:17:47,440 --> 00:17:52,400
so that's how we can see you know

441
00:17:49,840 --> 00:17:54,240
what the tuple representation of the

442
00:17:52,400 --> 00:17:55,840
above function is

443
00:17:54,240 --> 00:17:57,440
uh

444
00:17:55,840 --> 00:17:59,280
and you know we can move on here's the

445
00:17:57,440 --> 00:18:01,039
go to example we saw right at the start

446
00:17:59,280 --> 00:18:03,200
that i thought was really cool

447
00:18:01,039 --> 00:18:04,720
um just take a little bit it doesn't

448
00:18:03,200 --> 00:18:06,480
matter too much it just works like a

449
00:18:04,720 --> 00:18:08,160
go-to right

450
00:18:06,480 --> 00:18:09,440
and here's the macro implementation it

451
00:18:08,160 --> 00:18:11,200
looks a little scary but it's really

452
00:18:09,440 --> 00:18:13,679
simple um

453
00:18:11,200 --> 00:18:15,360
and so what it does is it checks if

454
00:18:13,679 --> 00:18:18,080
there is a string

455
00:18:15,360 --> 00:18:20,400
which has goto in it

456
00:18:18,080 --> 00:18:23,039
and then if it does it replaces the

457
00:18:20,400 --> 00:18:24,880
string loading command with

458
00:18:23,039 --> 00:18:27,039
a command that

459
00:18:24,880 --> 00:18:28,880
a jump command here with the wherever

460
00:18:27,039 --> 00:18:30,720
the go to was targeting

461
00:18:28,880 --> 00:18:32,400
and if there's a string

462
00:18:30,720 --> 00:18:34,720
which has a

463
00:18:32,400 --> 00:18:36,720
colon at the end or like a label then it

464
00:18:34,720 --> 00:18:38,640
replaces that instruction with a label

465
00:18:36,720 --> 00:18:41,520
instruction

466
00:18:38,640 --> 00:18:44,080
with whatever the label was uh and then

467
00:18:41,520 --> 00:18:46,240
otherwise it's just you know appends

468
00:18:44,080 --> 00:18:48,880
through normally and that's how the go

469
00:18:46,240 --> 00:18:51,360
to macro works i think it works so

470
00:18:48,880 --> 00:18:53,039
simply because there i added um

471
00:18:51,360 --> 00:18:55,679
because there is

472
00:18:53,039 --> 00:18:58,000
jump commands in

473
00:18:55,679 --> 00:19:00,160
uh the original bytecode and it is worth

474
00:18:58,000 --> 00:19:01,919
mentioning that this isn't an exact one

475
00:19:00,160 --> 00:19:03,600
to one of the original by code in human

476
00:19:01,919 --> 00:19:04,640
readable form i've made a few small

477
00:19:03,600 --> 00:19:07,440
changes

478
00:19:04,640 --> 00:19:09,440
just to make it better um but yeah it's

479
00:19:07,440 --> 00:19:10,799
pretty cool i think

480
00:19:09,440 --> 00:19:12,160
and then here's the last one which is a

481
00:19:10,799 --> 00:19:14,400
little complicated it took me a while to

482
00:19:12,160 --> 00:19:16,720
get my head around after i wrote it uh

483
00:19:14,400 --> 00:19:18,640
what it does it is just a

484
00:19:16,720 --> 00:19:20,559
an example to try to show that it's got

485
00:19:18,640 --> 00:19:22,080
so much power and you can really do

486
00:19:20,559 --> 00:19:23,360
anything you want with this you don't

487
00:19:22,080 --> 00:19:26,000
have to do something that already exists

488
00:19:23,360 --> 00:19:28,400
you can let your mind run three so to

489
00:19:26,000 --> 00:19:31,200
speak um and so heads one works is

490
00:19:28,400 --> 00:19:33,039
within the join and block

491
00:19:31,200 --> 00:19:35,919
it it takes the output of the previous

492
00:19:33,039 --> 00:19:37,200
line and feeds it in as the first input

493
00:19:35,919 --> 00:19:40,080
on the

494
00:19:37,200 --> 00:19:41,440
of the function on the next line um and

495
00:19:40,080 --> 00:19:42,960
and n equals three doesn't return

496
00:19:41,440 --> 00:19:45,039
anything in python

497
00:19:42,960 --> 00:19:47,440
uh but i've made it return like a walrus

498
00:19:45,039 --> 00:19:48,240
operator just to make it uh just make

499
00:19:47,440 --> 00:19:50,480
this

500
00:19:48,240 --> 00:19:52,240
this whole thing work but so for this

501
00:19:50,480 --> 00:19:54,320
example right it goes from range three

502
00:19:52,240 --> 00:19:55,440
to six and then it sums that range from

503
00:19:54,320 --> 00:19:57,840
three to six

504
00:19:55,440 --> 00:19:59,919
and then it times it by two and then it

505
00:19:57,840 --> 00:20:01,120
prints that output and which does print

506
00:19:59,919 --> 00:20:02,880
twenty four and then you can see the

507
00:20:01,120 --> 00:20:03,840
maths just here

508
00:20:02,880 --> 00:20:05,200
um

509
00:20:03,840 --> 00:20:07,200
and i think that's really cool this is

510
00:20:05,200 --> 00:20:08,559
just you know i think that's it's pretty

511
00:20:07,200 --> 00:20:09,280
crazy

512
00:20:08,559 --> 00:20:11,440
so

513
00:20:09,280 --> 00:20:14,080
uh just miscellaneous

514
00:20:11,440 --> 00:20:16,320
should you use this in your code um i i

515
00:20:14,080 --> 00:20:17,760
would i hope you wouldn't i think it's

516
00:20:16,320 --> 00:20:19,440
you know it's really bad meta

517
00:20:17,760 --> 00:20:21,280
programming a lot of times

518
00:20:19,440 --> 00:20:22,720
not optimal to using your code at all it

519
00:20:21,280 --> 00:20:24,159
can lead to some very hard to read crow

520
00:20:22,720 --> 00:20:26,799
which we see before

521
00:20:24,159 --> 00:20:30,400
um and i think fundamentally

522
00:20:26,799 --> 00:20:32,080
it's it's kind of anti-pythonic it's not

523
00:20:30,400 --> 00:20:34,880
it's it's against the principles of

524
00:20:32,080 --> 00:20:36,080
python um and so i've got the zener

525
00:20:34,880 --> 00:20:39,120
python here

526
00:20:36,080 --> 00:20:42,159
the zen of python the python poem that

527
00:20:39,120 --> 00:20:44,720
explains exactly

528
00:20:42,159 --> 00:20:47,760
uh the the core design principles of

529
00:20:44,720 --> 00:20:50,720
python and i think you can see that it's

530
00:20:47,760 --> 00:20:52,400
it's just against most of it right um

531
00:20:50,720 --> 00:20:54,559
you know explicit is better than

532
00:20:52,400 --> 00:20:57,440
implicit that's that's the whole idea

533
00:20:54,559 --> 00:21:00,799
isn't it right to try to build

534
00:20:57,440 --> 00:21:02,400
uh implicit changes to python

535
00:21:00,799 --> 00:21:04,159
and then you know

536
00:21:02,400 --> 00:21:05,919
what's the other one oh yeah

537
00:21:04,159 --> 00:21:07,840
if the implementation is hard to explain

538
00:21:05,919 --> 00:21:09,280
it's a bad idea i think that's that's a

539
00:21:07,840 --> 00:21:10,559
given right i spent a whole 10 minutes

540
00:21:09,280 --> 00:21:13,200
trying to explain it

541
00:21:10,559 --> 00:21:14,960
um and then the last one being special

542
00:21:13,200 --> 00:21:16,320
cases aren't special enough to break the

543
00:21:14,960 --> 00:21:18,400
rules right that's the whole point we're

544
00:21:16,320 --> 00:21:20,640
trying to break the rules of python in

545
00:21:18,400 --> 00:21:22,480
order to account for these special cases

546
00:21:20,640 --> 00:21:24,960
right um

547
00:21:22,480 --> 00:21:27,440
and yeah so then you're probably asking

548
00:21:24,960 --> 00:21:29,360
why did you do this then you know what's

549
00:21:27,440 --> 00:21:30,880
what's the whole point and i think you

550
00:21:29,360 --> 00:21:32,559
know you can't do the stuff that list

551
00:21:30,880 --> 00:21:34,480
macros allow you to do you can extend

552
00:21:32,559 --> 00:21:36,559
python syntax you can even write domain

553
00:21:34,480 --> 00:21:38,400
specific languages if you really want to

554
00:21:36,559 --> 00:21:39,440
there's some element of code reuses but

555
00:21:38,400 --> 00:21:41,280
i think

556
00:21:39,440 --> 00:21:43,280
at the end of the day it's just a lot of

557
00:21:41,280 --> 00:21:44,720
fun it was fun to make and i think it's

558
00:21:43,280 --> 00:21:45,919
fun to mess around with the internals of

559
00:21:44,720 --> 00:21:47,360
python

560
00:21:45,919 --> 00:21:49,440
and it's fun to extend language like

561
00:21:47,360 --> 00:21:51,120
this i think it's you know i i enjoyed i

562
00:21:49,440 --> 00:21:52,559
enjoyed doing it and you know i hope you

563
00:21:51,120 --> 00:21:55,120
would enjoy messing around with the

564
00:21:52,559 --> 00:21:57,280
library a little bit uh and yeah so

565
00:21:55,120 --> 00:21:59,200
that's the end of my talk here are the

566
00:21:57,280 --> 00:22:01,440
links around the biker so if you want to

567
00:21:59,200 --> 00:22:02,559
check out the github you can here's a

568
00:22:01,440 --> 00:22:04,159
link

569
00:22:02,559 --> 00:22:07,039
uh and if you want to just install the

570
00:22:04,159 --> 00:22:08,480
library go ahead uh you know just do it

571
00:22:07,039 --> 00:22:09,520
instantly like this

572
00:22:08,480 --> 00:22:12,799
there's the

573
00:22:09,520 --> 00:22:14,159
installation and command is on pip

574
00:22:12,799 --> 00:22:16,000
but i would recommend reading through

575
00:22:14,159 --> 00:22:17,679
the documentation on github if you do

576
00:22:16,000 --> 00:22:18,799
want to give this library a little bit

577
00:22:17,679 --> 00:22:20,400
of a try

578
00:22:18,799 --> 00:22:22,960
uh thank you

579
00:22:20,400 --> 00:22:25,440
fantastic thanks so much

580
00:22:22,960 --> 00:22:27,760
our next talk is a group talk we'll be

581
00:22:25,440 --> 00:22:30,080
hearing from emilia dahlia vyan and

582
00:22:27,760 --> 00:22:31,919
elizabeth about building cooker berry

583
00:22:30,080 --> 00:22:33,600
apps for learning and they've been

584
00:22:31,919 --> 00:22:34,720
programming just since the beginning of

585
00:22:33,600 --> 00:22:36,799
this year

586
00:22:34,720 --> 00:22:39,520
when i asked their favorite emoji it was

587
00:22:36,799 --> 00:22:42,960
a 50 50 split between the rolling eyes

588
00:22:39,520 --> 00:22:45,039
emoji and the sun emoji a dangerous mix

589
00:22:42,960 --> 00:22:46,480
i hope you've got sunnies too

590
00:22:45,039 --> 00:22:48,080
when i asked what they recently learned

591
00:22:46,480 --> 00:22:50,320
about computing they said they've been

592
00:22:48,080 --> 00:22:52,880
learning about caesar ciphers encoding

593
00:22:50,320 --> 00:22:55,520
and decoding and also about the general

594
00:22:52,880 --> 00:22:56,799
concept of computers not having a human

595
00:22:55,520 --> 00:22:59,200
language

596
00:22:56,799 --> 00:23:00,720
it's a deep rabbit hole to go down but

597
00:22:59,200 --> 00:23:03,520
an interesting one

598
00:23:00,720 --> 00:23:03,520
let's hear their talk

599
00:23:03,600 --> 00:23:08,320
hi i'm amelia

600
00:23:05,520 --> 00:23:10,000
elizabeth

601
00:23:08,320 --> 00:23:12,640
dalia

602
00:23:10,000 --> 00:23:14,080
and we're in u9 at plc sydney

603
00:23:12,640 --> 00:23:15,760
and we are the cambridge computer

604
00:23:14,080 --> 00:23:18,240
science class

605
00:23:15,760 --> 00:23:20,000
i've been coding for about seven months

606
00:23:18,240 --> 00:23:22,400
started this year

607
00:23:20,000 --> 00:23:24,080
since the start of this year around

608
00:23:22,400 --> 00:23:25,919
seven months

609
00:23:24,080 --> 00:23:27,840
i got involved in computer science

610
00:23:25,919 --> 00:23:29,600
because i believe that if i learn about

611
00:23:27,840 --> 00:23:31,679
something that keeps evolving it will

612
00:23:29,600 --> 00:23:33,120
help me in the future

613
00:23:31,679 --> 00:23:34,320
i think it is important to know about

614
00:23:33,120 --> 00:23:37,120
because the time we are living in

615
00:23:34,320 --> 00:23:38,640
revolves more and more around computers

616
00:23:37,120 --> 00:23:40,960
i believe that computer science will be

617
00:23:38,640 --> 00:23:42,559
a helpful tool in our future

618
00:23:40,960 --> 00:23:44,559
and i thought it would be interesting to

619
00:23:42,559 --> 00:23:46,559
learn

620
00:23:44,559 --> 00:23:48,080
for our assignment our teacher ms noble

621
00:23:46,559 --> 00:23:49,600
challenged us to design and code

622
00:23:48,080 --> 00:23:50,960
programs in python that will help with

623
00:23:49,600 --> 00:23:52,720
learning

624
00:23:50,960 --> 00:23:55,120
to complete the challenge we got to use

625
00:23:52,720 --> 00:23:56,720
cooker berries an embedded device coded

626
00:23:55,120 --> 00:23:58,799
in micro python

627
00:23:56,720 --> 00:24:01,520
they have lots of inputs and outputs

628
00:23:58,799 --> 00:24:04,720
like buttons screens and speakers that

629
00:24:01,520 --> 00:24:06,400
we could use to create our programs

630
00:24:04,720 --> 00:24:08,400
together we came up with seven apps to

631
00:24:06,400 --> 00:24:12,159
help with practicing music and studying

632
00:24:08,400 --> 00:24:12,159
for school that we'll show you today

633
00:24:14,960 --> 00:24:18,720
i created two apps to help with language

634
00:24:16,640 --> 00:24:20,080
learning and organization i chose to

635
00:24:18,720 --> 00:24:21,440
create programs for this area of

636
00:24:20,080 --> 00:24:22,880
learning because i personally struggled

637
00:24:21,440 --> 00:24:24,960
with organizing myself and finding

638
00:24:22,880 --> 00:24:26,320
myself wanting to study i started by

639
00:24:24,960 --> 00:24:27,440
thinking about what i would want in a

640
00:24:26,320 --> 00:24:28,720
learning app and thought it would be

641
00:24:27,440 --> 00:24:30,720
good if i could use technology to

642
00:24:28,720 --> 00:24:33,120
emulate that so i decided to make my

643
00:24:30,720 --> 00:24:34,799
calendar and flashcard cookieberry apps

644
00:24:33,120 --> 00:24:36,480
my first app was my calendar app and i

645
00:24:34,799 --> 00:24:38,559
started off by working on the design and

646
00:24:36,480 --> 00:24:40,080
features i wanted my original idea was

647
00:24:38,559 --> 00:24:41,679
to have a large timetable template

648
00:24:40,080 --> 00:24:43,520
similar to a school timetable with all

649
00:24:41,679 --> 00:24:44,799
the days of the week however i realized

650
00:24:43,520 --> 00:24:46,320
this would not be possible since the

651
00:24:44,799 --> 00:24:48,240
cookberry pixels are quite large and

652
00:24:46,320 --> 00:24:49,600
there would not be enough space i came

653
00:24:48,240 --> 00:24:51,600
up with a new design that fit the screen

654
00:24:49,600 --> 00:24:53,919
which only showed one day at a time once

655
00:24:51,600 --> 00:24:55,440
i had my design i needed to start coding

656
00:24:53,919 --> 00:24:57,520
i first coded the lines that would make

657
00:24:55,440 --> 00:24:58,799
up the timetable boxes there's a lot of

658
00:24:57,520 --> 00:25:00,159
python to learn because this is the

659
00:24:58,799 --> 00:25:01,840
first time i'd encountered code for

660
00:25:00,159 --> 00:25:03,760
displaying boxes and shapes on the kooky

661
00:25:01,840 --> 00:25:05,440
berry getting it to look like my design

662
00:25:03,760 --> 00:25:06,720
took a lot of trial and error i also had

663
00:25:05,440 --> 00:25:07,919
a lot of lists of data to organize to

664
00:25:06,720 --> 00:25:10,159
ensure that different study sessions

665
00:25:07,919 --> 00:25:11,679
would associate with their day

666
00:25:10,159 --> 00:25:13,279
i also needed to do things like get the

667
00:25:11,679 --> 00:25:14,880
week to loop back round to the start

668
00:25:13,279 --> 00:25:17,200
when using buttons to click the days in

669
00:25:14,880 --> 00:25:18,960
the timetable i needed to use a special

670
00:25:17,200 --> 00:25:20,480
case when i reached the end of the week

671
00:25:18,960 --> 00:25:22,240
so i could get back to the data stored

672
00:25:20,480 --> 00:25:24,080
in my list for the start of the week

673
00:25:22,240 --> 00:25:26,240
when i run my app it opens to sunday and

674
00:25:24,080 --> 00:25:29,200
the task that day to go the next day

675
00:25:26,240 --> 00:25:30,640
click d to go the previous day click c

676
00:25:29,200 --> 00:25:32,080
for my second app i thought about how

677
00:25:30,640 --> 00:25:33,760
flashcards are helpful tool when

678
00:25:32,080 --> 00:25:34,720
learning so i thought a flashcard app

679
00:25:33,760 --> 00:25:37,200
would be a good choice for this

680
00:25:34,720 --> 00:25:38,559
assignment i wanted to be able to show a

681
00:25:37,200 --> 00:25:40,159
word in one language and then have the

682
00:25:38,559 --> 00:25:42,640
user click to find out and memorize the

683
00:25:40,159 --> 00:25:44,240
translation it ended up working so if a

684
00:25:42,640 --> 00:25:46,240
is pressed it will display a word to

685
00:25:44,240 --> 00:25:48,480
memorize in this case i chose various

686
00:25:46,240 --> 00:25:51,279
french words if b is pressed the english

687
00:25:48,480 --> 00:25:52,960
is shown if d is pressed the next card

688
00:25:51,279 --> 00:25:54,640
is shown with the english option i

689
00:25:52,960 --> 00:25:56,240
wanted to make sure that it would be

690
00:25:54,640 --> 00:25:57,679
able to go backwards through the cards

691
00:25:56,240 --> 00:25:59,600
as well but i couldn't figure out how

692
00:25:57,679 --> 00:26:01,120
with the time i had when the final words

693
00:25:59,600 --> 00:26:02,880
are reached and d is pressed to go to

694
00:26:01,120 --> 00:26:04,480
the next one a congratulatory message

695
00:26:02,880 --> 00:26:06,559
appears as well as the vocab that should

696
00:26:04,480 --> 00:26:09,520
be memorized the user has the option to

697
00:26:06,559 --> 00:26:11,039
repeat the system by clicking a

698
00:26:09,520 --> 00:26:12,240
the hardest part of this app was finding

699
00:26:11,039 --> 00:26:13,679
a way to switch between the cards

700
00:26:12,240 --> 00:26:15,520
because there were multiple bits of data

701
00:26:13,679 --> 00:26:17,760
associated with each screen

702
00:26:15,520 --> 00:26:19,360
i really enjoyed learning about these

703
00:26:17,760 --> 00:26:21,039
new ways to use python because i never

704
00:26:19,360 --> 00:26:22,480
encountered them before

705
00:26:21,039 --> 00:26:24,320
in the future i hope to utilize what

706
00:26:22,480 --> 00:26:27,600
i've learned so i can find ways to

707
00:26:24,320 --> 00:26:27,600
improve my programming skills

708
00:26:29,440 --> 00:26:33,919
i created an app to help with studying

709
00:26:32,159 --> 00:26:35,919
i chose to create a program for this

710
00:26:33,919 --> 00:26:37,279
area of learning because it helps lots

711
00:26:35,919 --> 00:26:40,159
of people when they are studying for

712
00:26:37,279 --> 00:26:42,080
multiple subjects at one given time

713
00:26:40,159 --> 00:26:44,080
i started out by thinking about how i

714
00:26:42,080 --> 00:26:45,840
could use programming and the features

715
00:26:44,080 --> 00:26:48,720
of the kookaberry like the buttons and

716
00:26:45,840 --> 00:26:51,919
screen to help students learn

717
00:26:48,720 --> 00:26:54,159
and so i decided to make a pop quiz app

718
00:26:51,919 --> 00:26:56,080
to do this i first had to come up with a

719
00:26:54,159 --> 00:26:58,320
plan of what questions to put in the

720
00:26:56,080 --> 00:27:01,440
quiz as well as coming up with both the

721
00:26:58,320 --> 00:27:04,159
correct and incorrect answers

722
00:27:01,440 --> 00:27:06,559
to code the actual quiz i had to use a

723
00:27:04,159 --> 00:27:08,880
lot of lists for storing the question

724
00:27:06,559 --> 00:27:10,400
the multiple choice answers as well as a

725
00:27:08,880 --> 00:27:12,640
list that says which is the correct

726
00:27:10,400 --> 00:27:14,480
answer for each question

727
00:27:12,640 --> 00:27:16,080
getting it to work was tricky because

728
00:27:14,480 --> 00:27:18,720
sometimes the question wouldn't match up

729
00:27:16,080 --> 00:27:20,559
with the corresponding answer but once i

730
00:27:18,720 --> 00:27:23,200
made sure my lists were in the right

731
00:27:20,559 --> 00:27:25,360
order of answers and questions and that

732
00:27:23,200 --> 00:27:26,640
i was using the right indexes it all

733
00:27:25,360 --> 00:27:28,640
worked

734
00:27:26,640 --> 00:27:31,360
along the way i changed the design of my

735
00:27:28,640 --> 00:27:32,960
project like including more questions

736
00:27:31,360 --> 00:27:34,799
since i thought this would be useful to

737
00:27:32,960 --> 00:27:37,440
give the user more questions to test

738
00:27:34,799 --> 00:27:37,440
themselves on

739
00:27:38,159 --> 00:27:42,640
in the end i got an app that works by

740
00:27:40,240 --> 00:27:44,880
pressing a to start it and then when the

741
00:27:42,640 --> 00:27:48,559
question and answer options appear you

742
00:27:44,880 --> 00:27:50,399
select an answer using buttons a to d

743
00:27:48,559 --> 00:27:52,159
once you have chosen your answer it

744
00:27:50,399 --> 00:27:54,799
tells you if you are correct or not and

745
00:27:52,159 --> 00:27:56,640
lets you proceed to the next question

746
00:27:54,799 --> 00:27:58,640
in the future i hope to code this app

747
00:27:56,640 --> 00:28:00,799
again so i can improve some features of

748
00:27:58,640 --> 00:28:03,120
it like more questions and maybe a score

749
00:28:00,799 --> 00:28:06,159
tally at the end and continue adding to

750
00:28:03,120 --> 00:28:06,159
it to make it better

751
00:28:10,640 --> 00:28:14,559
to do this project i started out by

752
00:28:12,640 --> 00:28:16,320
thinking about what activities i do for

753
00:28:14,559 --> 00:28:18,640
school and who could help me with my own

754
00:28:16,320 --> 00:28:20,880
learning i played musical instrument in

755
00:28:18,640 --> 00:28:22,640
school so i decided to make an app that

756
00:28:20,880 --> 00:28:23,840
would make it easier to practice music

757
00:28:22,640 --> 00:28:26,000
from home

758
00:28:23,840 --> 00:28:28,240
so i decided to make two apps a

759
00:28:26,000 --> 00:28:29,919
metronome that helps you play in time

760
00:28:28,240 --> 00:28:32,240
and an app that keeps track of how many

761
00:28:29,919 --> 00:28:34,159
times you've practiced different songs

762
00:28:32,240 --> 00:28:35,760
for my metronome app i used a cooker

763
00:28:34,159 --> 00:28:38,080
berry with the speaker module to play

764
00:28:35,760 --> 00:28:40,399
the beat to code my app i had to think

765
00:28:38,080 --> 00:28:42,159
about how people would interact with it

766
00:28:40,399 --> 00:28:44,480
and i decided to use four buttons on the

767
00:28:42,159 --> 00:28:47,200
cooker berry for different functions on

768
00:28:44,480 --> 00:28:49,360
off faster and slower

769
00:28:47,200 --> 00:28:51,600
to reuse the metronome you click a to

770
00:28:49,360 --> 00:28:53,919
start it once a is clicked the beat will

771
00:28:51,600 --> 00:28:56,240
start coming from the speaker if you

772
00:28:53,919 --> 00:28:58,240
click c the tempo will become slower and

773
00:28:56,240 --> 00:29:02,480
if you click d the tempo will become

774
00:28:58,240 --> 00:29:02,480
faster to stop the metronome click b

775
00:29:03,279 --> 00:29:08,240
because i wanted to be able to vary the

776
00:29:05,440 --> 00:29:10,240
speed of my metronome gaining the timing

777
00:29:08,240 --> 00:29:12,480
was tricky i had to have two different

778
00:29:10,240 --> 00:29:14,240
time values so that it would stay in b

779
00:29:12,480 --> 00:29:16,399
for the length of the b and the time

780
00:29:14,240 --> 00:29:18,640
between beats but once i understood how

781
00:29:16,399 --> 00:29:20,640
to get the time right i could add two or

782
00:29:18,640 --> 00:29:23,520
subtract from the variables to change

783
00:29:20,640 --> 00:29:25,679
the tempo to get it working

784
00:29:23,520 --> 00:29:27,039
once i had my metronome working i wanted

785
00:29:25,679 --> 00:29:29,279
a way to track the songs i was

786
00:29:27,039 --> 00:29:31,039
practicing i started out by coding a

787
00:29:29,279 --> 00:29:33,760
list of all the names of the songs i

788
00:29:31,039 --> 00:29:35,919
wanted to track to practice on

789
00:29:33,760 --> 00:29:37,679
once i had done that i created another

790
00:29:35,919 --> 00:29:39,840
list that would keep a tally for each

791
00:29:37,679 --> 00:29:42,559
song so that we would have a different

792
00:29:39,840 --> 00:29:44,000
count for each songs in the first list

793
00:29:42,559 --> 00:29:45,919
getting the tally to match up with

794
00:29:44,000 --> 00:29:48,080
correct song was tricky because it would

795
00:29:45,919 --> 00:29:50,240
keep the same tally but then go to the

796
00:29:48,080 --> 00:29:53,120
next song but once i worked out the

797
00:29:50,240 --> 00:29:55,440
indexes i got it working

798
00:29:53,120 --> 00:29:58,159
my finished app looked like this

799
00:29:55,440 --> 00:30:00,240
you can click a and b to move backwards

800
00:29:58,159 --> 00:30:02,880
and forwards between different songs and

801
00:30:00,240 --> 00:30:05,520
click c and d to add or

802
00:30:02,880 --> 00:30:07,279
take away from the tally i even added an

803
00:30:05,520 --> 00:30:09,840
extra feature where once the tally

804
00:30:07,279 --> 00:30:11,760
reaches 5 on the song this neopixel

805
00:30:09,840 --> 00:30:14,960
light display will flash to congratulate

806
00:30:11,760 --> 00:30:17,120
you for you reaching your goal

807
00:30:14,960 --> 00:30:19,120
i really enjoyed making my apps because

808
00:30:17,120 --> 00:30:21,120
i could put my skills to use and create

809
00:30:19,120 --> 00:30:23,840
something of my own rather than copying

810
00:30:21,120 --> 00:30:23,840
from a booklet

811
00:30:27,440 --> 00:30:31,279
i created two wraps to help with study i

812
00:30:29,600 --> 00:30:32,880
chose to create programs for this area

813
00:30:31,279 --> 00:30:34,799
of learning because i do not have a good

814
00:30:32,880 --> 00:30:36,960
study routine and never feel motivated

815
00:30:34,799 --> 00:30:38,720
to study

816
00:30:36,960 --> 00:30:40,159
i started out by thinking about the ways

817
00:30:38,720 --> 00:30:42,000
that have helped me to study and thought

818
00:30:40,159 --> 00:30:45,120
it'd be good if i could use technology

819
00:30:42,000 --> 00:30:47,200
to make study time in quota so first i

820
00:30:45,120 --> 00:30:49,279
decided to make an automatic pomodoro

821
00:30:47,200 --> 00:30:51,679
method timer which is a method that is

822
00:30:49,279 --> 00:30:53,919
25 minutes of study and a five minute

823
00:30:51,679 --> 00:30:55,840
break and this continues

824
00:30:53,919 --> 00:30:58,080
and once i've done that i made an app

825
00:30:55,840 --> 00:30:59,840
that displays inspirational quotes to

826
00:30:58,080 --> 00:31:01,519
keep me motivated

827
00:30:59,840 --> 00:31:03,760
the first thing i did was work out how

828
00:31:01,519 --> 00:31:05,760
my study time should work my first

829
00:31:03,760 --> 00:31:07,519
version of the project was an app that

830
00:31:05,760 --> 00:31:09,200
would start when you press button a and

831
00:31:07,519 --> 00:31:11,279
it would start counting down from 25

832
00:31:09,200 --> 00:31:13,039
minutes for the study session and once

833
00:31:11,279 --> 00:31:15,519
that finished it would count down from

834
00:31:13,039 --> 00:31:17,600
five minutes for the break time

835
00:31:15,519 --> 00:31:19,919
i then added on extra features like

836
00:31:17,600 --> 00:31:21,519
doing multiple sessions in a row adding

837
00:31:19,919 --> 00:31:24,320
a speaker to make sounds at the end of

838
00:31:21,519 --> 00:31:26,080
each set each session the ability to end

839
00:31:24,320 --> 00:31:28,240
the session and account for how many

840
00:31:26,080 --> 00:31:30,000
sessions you've done in a row

841
00:31:28,240 --> 00:31:32,159
the trickiest thing about making the

842
00:31:30,000 --> 00:31:34,159
time was that the coco really long by

843
00:31:32,159 --> 00:31:35,760
the end and i needed to keep adding to

844
00:31:34,159 --> 00:31:37,679
and editing lots of different parts of

845
00:31:35,760 --> 00:31:39,760
the code to create my new features

846
00:31:37,679 --> 00:31:41,679
because it was all connected but with

847
00:31:39,760 --> 00:31:42,880
the help of my teacher i managed to work

848
00:31:41,679 --> 00:31:43,840
out how to code it all and get it

849
00:31:42,880 --> 00:31:45,919
working

850
00:31:43,840 --> 00:31:47,679
my motivational quote was much more

851
00:31:45,919 --> 00:31:49,919
straightforward to make it wasn't

852
00:31:47,679 --> 00:31:51,919
particularly hard to code since i just

853
00:31:49,919 --> 00:31:56,720
started to learn how to code it was

854
00:31:51,919 --> 00:31:56,720
tricky to locate and fix syntax errors

855
00:31:56,799 --> 00:32:00,880
using my app is also straightforward you

856
00:31:58,799 --> 00:32:02,720
can just press button a this will make a

857
00:32:00,880 --> 00:32:04,000
random code display on the screen of the

858
00:32:02,720 --> 00:32:05,679
quicker berry

859
00:32:04,000 --> 00:32:08,559
you can press button every time you want

860
00:32:05,679 --> 00:32:10,559
a new quote to read

861
00:32:08,559 --> 00:32:12,159
i really enjoy creating my own apps

862
00:32:10,559 --> 00:32:15,120
because i was able to code without a

863
00:32:12,159 --> 00:32:17,120
step-by-step guide in future i hope to

864
00:32:15,120 --> 00:32:18,720
code more apps again so i can implement

865
00:32:17,120 --> 00:32:20,799
what i've learned from this from the

866
00:32:18,720 --> 00:32:24,000
beginning of this year up until now and

867
00:32:20,799 --> 00:32:24,000
create even better apps

868
00:32:26,720 --> 00:32:30,320
i really enjoyed making these apps

869
00:32:28,559 --> 00:32:32,320
because it helped me understand what

870
00:32:30,320 --> 00:32:34,240
goes on inside a computer when using a

871
00:32:32,320 --> 00:32:36,159
similar app

872
00:32:34,240 --> 00:32:38,000
using python for this project was good

873
00:32:36,159 --> 00:32:39,679
because i learned most of the basics of

874
00:32:38,000 --> 00:32:41,360
python at the beginning of the year so i

875
00:32:39,679 --> 00:32:43,679
was able to understand the layout of the

876
00:32:41,360 --> 00:32:45,679
code and how to code my apps with cooker

877
00:32:43,679 --> 00:32:47,279
berry

878
00:32:45,679 --> 00:32:48,960
in the future i hope to use the cooker

879
00:32:47,279 --> 00:32:51,200
barrier again so i can create more apps

880
00:32:48,960 --> 00:32:53,200
with better coding

881
00:32:51,200 --> 00:32:54,960
using python for these products was good

882
00:32:53,200 --> 00:32:56,720
because we had spent the year learning

883
00:32:54,960 --> 00:32:58,399
about python which is relatively easy

884
00:32:56,720 --> 00:32:59,919
for a beginner and i look forward to

885
00:32:58,399 --> 00:33:02,320
creating more projects with python in

886
00:32:59,919 --> 00:33:02,320
the future

887
00:33:06,080 --> 00:33:11,200
that was fantastic consoles look out

888
00:33:09,039 --> 00:33:13,200
there are new games in town and here's

889
00:33:11,200 --> 00:33:14,880
another one to talk about

890
00:33:13,200 --> 00:33:17,279
samuel is talking about making an

891
00:33:14,880 --> 00:33:19,200
isometric game from scratch in python

892
00:33:17,279 --> 00:33:21,039
using pygame

893
00:33:19,200 --> 00:33:24,159
samuel's been programming for about six

894
00:33:21,039 --> 00:33:25,919
years so since year six and his favorite

895
00:33:24,159 --> 00:33:29,039
emoji is the

896
00:33:25,919 --> 00:33:30,720
glasses emoji glasses represent

897
00:33:29,039 --> 00:33:32,320
when i asked what he's recently learned

898
00:33:30,720 --> 00:33:34,159
about computers he replied that

899
00:33:32,320 --> 00:33:36,720
computers are bad at maths with

900
00:33:34,159 --> 00:33:39,279
fractions floating point errors occur

901
00:33:36,720 --> 00:33:41,760
when computers round numbers and they

902
00:33:39,279 --> 00:33:43,360
are quite scary these errors when they

903
00:33:41,760 --> 00:33:45,919
add up

904
00:33:43,360 --> 00:33:47,519
let's hear more about his talk

905
00:33:45,919 --> 00:33:49,760
good afternoon

906
00:33:47,519 --> 00:33:52,559
welcome to my talk on how i implemented

907
00:33:49,760 --> 00:33:54,240
isometric 3d in python's pie game

908
00:33:52,559 --> 00:33:55,840
by way of introductions my name is

909
00:33:54,240 --> 00:33:57,440
samuel price i go to san andreas

910
00:33:55,840 --> 00:34:00,080
cathedral school and i've been

911
00:33:57,440 --> 00:34:02,480
programming since grade six so about six

912
00:34:00,080 --> 00:34:04,000
to six years now and in those six years

913
00:34:02,480 --> 00:34:07,200
something i've always really wanted to

914
00:34:04,000 --> 00:34:08,800
do was develop a system for 3d

915
00:34:07,200 --> 00:34:10,320
i've very much always been able to

916
00:34:08,800 --> 00:34:12,240
believe that the best way to learn about

917
00:34:10,320 --> 00:34:13,520
something is to do it yourself i started

918
00:34:12,240 --> 00:34:16,000
off coding

919
00:34:13,520 --> 00:34:17,679
some text-based rpgs moved on to 2d

920
00:34:16,000 --> 00:34:19,280
games and i

921
00:34:17,679 --> 00:34:20,560
made a clone of

922
00:34:19,280 --> 00:34:23,599
quite a few

923
00:34:20,560 --> 00:34:26,159
uh 2d games if you can think of one

924
00:34:23,599 --> 00:34:28,960
i've probably done it

925
00:34:26,159 --> 00:34:31,359
but one day i really really wanted to do

926
00:34:28,960 --> 00:34:32,560
something in 3d the best system i could

927
00:34:31,359 --> 00:34:35,839
think of

928
00:34:32,560 --> 00:34:37,919
for me to implement was isometric 3d

929
00:34:35,839 --> 00:34:40,320
now if you remember your math math

930
00:34:37,919 --> 00:34:42,399
classes from grades one and two

931
00:34:40,320 --> 00:34:44,240
what you get is these little series of

932
00:34:42,399 --> 00:34:46,720
dots and you can connect them up to make

933
00:34:44,240 --> 00:34:49,040
shapes um essentially when it works if

934
00:34:46,720 --> 00:34:52,079
you want to travel say one unit in the x

935
00:34:49,040 --> 00:34:54,639
axis you draw one of these lines in this

936
00:34:52,079 --> 00:34:56,720
direction one of the y you do one of

937
00:34:54,639 --> 00:34:57,680
these lines here one with z one of those

938
00:34:56,720 --> 00:34:59,280
ones

939
00:34:57,680 --> 00:35:00,880
and you train those together to get

940
00:34:59,280 --> 00:35:02,800
different points and to make two

941
00:35:00,880 --> 00:35:04,800
different shapes

942
00:35:02,800 --> 00:35:06,880
which all works pretty well and you can

943
00:35:04,800 --> 00:35:08,079
see here that this is an implementation

944
00:35:06,880 --> 00:35:10,400
of that model

945
00:35:08,079 --> 00:35:11,520
it's just a little c where the ripples

946
00:35:10,400 --> 00:35:13,359
come out

947
00:35:11,520 --> 00:35:14,960
and you can you can see the depth you

948
00:35:13,359 --> 00:35:16,640
can understand how it functions in three

949
00:35:14,960 --> 00:35:17,760
dimensions you can see the rotation as

950
00:35:16,640 --> 00:35:20,480
well

951
00:35:17,760 --> 00:35:23,520
however the depth is also one problem of

952
00:35:20,480 --> 00:35:26,400
this see in orthographic 3d which is

953
00:35:23,520 --> 00:35:27,920
what we see in real life perspective 3d

954
00:35:26,400 --> 00:35:29,680
when something's further away you see a

955
00:35:27,920 --> 00:35:30,480
smaller when it's closer you see it is

956
00:35:29,680 --> 00:35:32,480
bigger

957
00:35:30,480 --> 00:35:34,640
but in isometric that's not quite how it

958
00:35:32,480 --> 00:35:36,640
works if you have just a series of dots

959
00:35:34,640 --> 00:35:38,160
it doesn't matter if you draw it on one

960
00:35:36,640 --> 00:35:40,400
place or another it's always going to be

961
00:35:38,160 --> 00:35:42,960
the same size

962
00:35:40,400 --> 00:35:45,200
and so you can see here in this game a

963
00:35:42,960 --> 00:35:47,359
small clone of temple run that i've

964
00:35:45,200 --> 00:35:48,320
found a way to overcome this

965
00:35:47,359 --> 00:35:50,640
see

966
00:35:48,320 --> 00:35:53,359
normally if you're jumping say from the

967
00:35:50,640 --> 00:35:54,800
left most lane be hard to tell if you're

968
00:35:53,359 --> 00:35:56,160
actually jumping from that lane or if

969
00:35:54,800 --> 00:35:58,160
you're just in another lane staying

970
00:35:56,160 --> 00:36:00,960
still

971
00:35:58,160 --> 00:36:03,680
which is why i implemented a shadow

972
00:36:00,960 --> 00:36:05,839
so you can see it shows under the player

973
00:36:03,680 --> 00:36:07,599
to add a sense of immersion and not

974
00:36:05,839 --> 00:36:10,240
break the suspension to disbelieve a

975
00:36:07,599 --> 00:36:12,000
player has when playing a game

976
00:36:10,240 --> 00:36:13,839
which is obviously quite useful for my

977
00:36:12,000 --> 00:36:15,119
purposes

978
00:36:13,839 --> 00:36:17,440
but

979
00:36:15,119 --> 00:36:19,359
another success and

980
00:36:17,440 --> 00:36:22,560
arguably another implementation of this

981
00:36:19,359 --> 00:36:22,560
model can be seen here

982
00:36:23,440 --> 00:36:27,520
now these shoot among you may notice

983
00:36:25,440 --> 00:36:28,640
that there is a slight problem with this

984
00:36:27,520 --> 00:36:30,079
clip

985
00:36:28,640 --> 00:36:32,960
and that is that

986
00:36:30,079 --> 00:36:35,040
it actually slightly overlaps in

987
00:36:32,960 --> 00:36:37,599
a few of the frames

988
00:36:35,040 --> 00:36:39,839
um which is to say that the shapes don't

989
00:36:37,599 --> 00:36:42,160
draw in the right order now thinking

990
00:36:39,839 --> 00:36:44,400
about draw or ordering

991
00:36:42,160 --> 00:36:48,160
this is the system i came up with to do

992
00:36:44,400 --> 00:36:51,040
it in isometric if you think uh you as

993
00:36:48,160 --> 00:36:52,800
the audience looking at me and this pen

994
00:36:51,040 --> 00:36:54,720
if you want to draw those to render in

995
00:36:52,800 --> 00:36:56,400
the correct order if i'm in the back you

996
00:36:54,720 --> 00:36:59,839
have to draw me first in a second so

997
00:36:56,400 --> 00:37:00,720
this shows over me rather than behind me

998
00:36:59,839 --> 00:37:03,280
and

999
00:37:00,720 --> 00:37:05,040
in order to do this i think of having a

1000
00:37:03,280 --> 00:37:06,560
line that goes straight from your vision

1001
00:37:05,040 --> 00:37:10,000
straight forward you don't have to care

1002
00:37:06,560 --> 00:37:13,359
about left or right or up or down

1003
00:37:10,000 --> 00:37:15,440
but if you look at me and then this pen

1004
00:37:13,359 --> 00:37:17,680
you think about this pen is actually

1005
00:37:15,440 --> 00:37:20,160
closer on that line going straight to

1006
00:37:17,680 --> 00:37:22,000
you than i am which is why this pen is

1007
00:37:20,160 --> 00:37:23,200
closer to you and should be rendered

1008
00:37:22,000 --> 00:37:25,200
last

1009
00:37:23,200 --> 00:37:26,960
and that's essentially what i did by

1010
00:37:25,200 --> 00:37:30,160
making a function that gives you the

1011
00:37:26,960 --> 00:37:32,160
normal to the to the uh camera for the

1012
00:37:30,160 --> 00:37:35,280
isometrics in 3d or for the

1013
00:37:32,160 --> 00:37:36,880
mathematically uninitiated of you and i

1014
00:37:35,280 --> 00:37:38,320
hope you look forward to that math i

1015
00:37:36,880 --> 00:37:40,320
certainly did

1016
00:37:38,320 --> 00:37:42,079
effectively just a line that goes

1017
00:37:40,320 --> 00:37:43,520
straight out from you like i was talking

1018
00:37:42,079 --> 00:37:47,599
about

1019
00:37:43,520 --> 00:37:49,520
i was able to sort shapes and points

1020
00:37:47,599 --> 00:37:50,720
to render in the correct order

1021
00:37:49,520 --> 00:37:53,040
however

1022
00:37:50,720 --> 00:37:55,280
in saying shapes it only really works

1023
00:37:53,040 --> 00:37:57,920
for points you can see in this example

1024
00:37:55,280 --> 00:37:59,200
here that this terrain that's generated

1025
00:37:57,920 --> 00:38:01,359
you can

1026
00:37:59,200 --> 00:38:02,880
see it drawn in the correct aura quarter

1027
00:38:01,359 --> 00:38:04,560
as the mountain moves closer to the

1028
00:38:02,880 --> 00:38:05,839
camera everything behind it is not

1029
00:38:04,560 --> 00:38:07,200
visible to you

1030
00:38:05,839 --> 00:38:09,520
and

1031
00:38:07,200 --> 00:38:11,359
that's exactly as you'd expect but then

1032
00:38:09,520 --> 00:38:13,040
why with those rectangular prisms does

1033
00:38:11,359 --> 00:38:16,079
it not always work

1034
00:38:13,040 --> 00:38:18,960
and as i understand it that's simply

1035
00:38:16,079 --> 00:38:21,599
a problem with discrete mathematics see

1036
00:38:18,960 --> 00:38:24,079
my formula is just built for working out

1037
00:38:21,599 --> 00:38:25,839
uh the position that it has in that line

1038
00:38:24,079 --> 00:38:27,599
which is only a point

1039
00:38:25,839 --> 00:38:28,720
not something that has width and height

1040
00:38:27,599 --> 00:38:30,400
and depth

1041
00:38:28,720 --> 00:38:31,920
um and that's definitely something i'm

1042
00:38:30,400 --> 00:38:33,520
going to be working on in the future to

1043
00:38:31,920 --> 00:38:34,640
try and overcome that or work a way

1044
00:38:33,520 --> 00:38:36,960
around it

1045
00:38:34,640 --> 00:38:39,599
but enough of these limitations

1046
00:38:36,960 --> 00:38:42,000
here's something i actually made with it

1047
00:38:39,599 --> 00:38:43,920
what you see here on the screen is my

1048
00:38:42,000 --> 00:38:45,599
year 12 major work for software design

1049
00:38:43,920 --> 00:38:48,240
and development

1050
00:38:45,599 --> 00:38:50,640
this is called ecosim as it is a

1051
00:38:48,240 --> 00:38:54,079
simulator of ecosystems you have plants

1052
00:38:50,640 --> 00:38:55,599
and animals interacting eating drinking

1053
00:38:54,079 --> 00:38:57,440
all that stuff

1054
00:38:55,599 --> 00:39:00,880
and the best way to think of to deliver

1055
00:38:57,440 --> 00:39:03,119
that was an isometric 3d to give it 2d

1056
00:39:00,880 --> 00:39:05,680
would mean have to flatten out animals

1057
00:39:03,119 --> 00:39:07,680
or have them represented by a character

1058
00:39:05,680 --> 00:39:09,520
i'm all that works i do value the

1059
00:39:07,680 --> 00:39:12,320
immersion even if it is only the

1060
00:39:09,520 --> 00:39:13,040
immersion offered by a voxelized model

1061
00:39:12,320 --> 00:39:16,880
like

1062
00:39:13,040 --> 00:39:16,880
minecraft or terrarium for example

1063
00:39:16,960 --> 00:39:21,200
but in doing this you can rotate the

1064
00:39:18,880 --> 00:39:22,640
models in 3d you can see them moving

1065
00:39:21,200 --> 00:39:24,079
around and even where they're going to

1066
00:39:22,640 --> 00:39:25,359
move as they point in that direction

1067
00:39:24,079 --> 00:39:27,920
when they walk

1068
00:39:25,359 --> 00:39:30,160
and you can notice the view ordering is

1069
00:39:27,920 --> 00:39:32,560
almost correct as when something walks

1070
00:39:30,160 --> 00:39:35,440
behind a tree it does actually go behind

1071
00:39:32,560 --> 00:39:38,240
it and in front of it when it's in front

1072
00:39:35,440 --> 00:39:40,720
which is definitely a success

1073
00:39:38,240 --> 00:39:42,960
however in saying that this did require

1074
00:39:40,720 --> 00:39:44,720
significant optimization

1075
00:39:42,960 --> 00:39:46,400
now for any of you that have tried to

1076
00:39:44,720 --> 00:39:48,880
make a rather significant project in

1077
00:39:46,400 --> 00:39:51,280
python probably well-known does not

1078
00:39:48,880 --> 00:39:54,079
always run the fastest

1079
00:39:51,280 --> 00:39:56,240
and a lot of that is definitely on me i

1080
00:39:54,079 --> 00:39:57,440
am not by any means the most experienced

1081
00:39:56,240 --> 00:39:58,560
programmer

1082
00:39:57,440 --> 00:40:00,640
um

1083
00:39:58,560 --> 00:40:03,119
and what i ended up coming up with was

1084
00:40:00,640 --> 00:40:06,079
two ways to optimize it the first way is

1085
00:40:03,119 --> 00:40:07,839
thinking of um as you see on the screen

1086
00:40:06,079 --> 00:40:09,440
here there are only certain tiles that

1087
00:40:07,839 --> 00:40:10,960
are actually visible to you you can

1088
00:40:09,440 --> 00:40:13,119
imagine that whole bunch of childs are

1089
00:40:10,960 --> 00:40:14,640
off screen that aren't and drawing those

1090
00:40:13,119 --> 00:40:17,200
is a waste of processing power

1091
00:40:14,640 --> 00:40:20,319
especially if you are trying to do that

1092
00:40:17,200 --> 00:40:22,880
60 times a second

1093
00:40:20,319 --> 00:40:26,480
so i i have done that by thinking of

1094
00:40:22,880 --> 00:40:28,160
each row as a line and doing some linear

1095
00:40:26,480 --> 00:40:30,160
algebra you can work out when the line

1096
00:40:28,160 --> 00:40:31,520
travels off the screen and that's the

1097
00:40:30,160 --> 00:40:33,680
wellington

1098
00:40:31,520 --> 00:40:35,440
but what if you were to zoom out you

1099
00:40:33,680 --> 00:40:37,760
would still be seeing everything and

1100
00:40:35,440 --> 00:40:40,079
optimization wouldn't help you there

1101
00:40:37,760 --> 00:40:42,000
because there's nothing off screen

1102
00:40:40,079 --> 00:40:45,359
so the next idea i had was just to

1103
00:40:42,000 --> 00:40:47,520
pre-render the sprite so turn a 3d map

1104
00:40:45,359 --> 00:40:49,599
into a 2d sprite which you can draw on

1105
00:40:47,520 --> 00:40:52,240
the screen each frame a lot more easily

1106
00:40:49,599 --> 00:40:53,760
than you can render every single tile

1107
00:40:52,240 --> 00:40:55,440
and that's what i ended up going with so

1108
00:40:53,760 --> 00:40:58,880
in this game there is a loading screen

1109
00:40:55,440 --> 00:41:01,280
as you open it up to pre-referendum

1110
00:40:58,880 --> 00:41:02,480
but moving into the future what i really

1111
00:41:01,280 --> 00:41:04,319
hope to do

1112
00:41:02,480 --> 00:41:05,599
is

1113
00:41:04,319 --> 00:41:08,480
to

1114
00:41:05,599 --> 00:41:10,400
kind of chunk these pre-renderings so to

1115
00:41:08,480 --> 00:41:14,720
pre-render every eight by eight for

1116
00:41:10,400 --> 00:41:17,920
example tile map and then uh

1117
00:41:14,720 --> 00:41:20,800
draw that as a texture-mapped

1118
00:41:17,920 --> 00:41:23,440
block of its own which i think is an

1119
00:41:20,800 --> 00:41:26,240
achievable goal for the future

1120
00:41:23,440 --> 00:41:28,480
although it may take me years if i am to

1121
00:41:26,240 --> 00:41:29,920
do it all myself but enough about all

1122
00:41:28,480 --> 00:41:31,680
that

1123
00:41:29,920 --> 00:41:33,680
for those among you that are interested

1124
00:41:31,680 --> 00:41:35,680
as i know i certainly would be

1125
00:41:33,680 --> 00:41:38,640
this is the mathematics for this this

1126
00:41:35,680 --> 00:41:40,000
transformation of turning um the real

1127
00:41:38,640 --> 00:41:41,680
in-game coordinates into drawing

1128
00:41:40,000 --> 00:41:43,359
coordinates as you can see on the screen

1129
00:41:41,680 --> 00:41:44,560
here

1130
00:41:43,359 --> 00:41:47,200
what you have

1131
00:41:44,560 --> 00:41:48,960
is first off you have your decode as

1132
00:41:47,200 --> 00:41:51,200
your drawing coordinates your arc as

1133
00:41:48,960 --> 00:41:53,359
your real or in-game coordinates add for

1134
00:41:51,200 --> 00:41:56,640
the camera offset or add

1135
00:41:53,359 --> 00:41:58,720
as in addition uh zoom for how zoomed in

1136
00:41:56,640 --> 00:42:00,720
the camera is

1137
00:41:58,720 --> 00:42:02,400
and then what you have is the real

1138
00:42:00,720 --> 00:42:04,480
coordinate

1139
00:42:02,400 --> 00:42:06,319
time the real x coordinate times sine of

1140
00:42:04,480 --> 00:42:08,079
theta plus the real life water times cos

1141
00:42:06,319 --> 00:42:11,200
of theta and you're talking about your

1142
00:42:08,079 --> 00:42:13,200
horizontal drawn coordinate and then for

1143
00:42:11,200 --> 00:42:14,480
the horizontal vertical component you

1144
00:42:13,200 --> 00:42:15,520
have

1145
00:42:14,480 --> 00:42:17,599
the real

1146
00:42:15,520 --> 00:42:21,280
in game x times cos of theta minus the

1147
00:42:17,599 --> 00:42:22,960
real y times sine of theta and you may

1148
00:42:21,280 --> 00:42:25,200
be wondering why i have a different sine

1149
00:42:22,960 --> 00:42:26,319
theta and closely or different lines and

1150
00:42:25,200 --> 00:42:27,839
to put it

1151
00:42:26,319 --> 00:42:29,680
more or less simply

1152
00:42:27,839 --> 00:42:31,520
that's basically a 90 degree rotation

1153
00:42:29,680 --> 00:42:33,760
because if you're looking straight down

1154
00:42:31,520 --> 00:42:35,040
a square is always supposed to have a

1155
00:42:33,760 --> 00:42:36,720
right angle

1156
00:42:35,040 --> 00:42:39,440
and i wanted to guarantee you that but

1157
00:42:36,720 --> 00:42:42,000
you might also be thinking well

1158
00:42:39,440 --> 00:42:44,079
what's this other cause and sign

1159
00:42:42,000 --> 00:42:45,119
and that

1160
00:42:44,079 --> 00:42:47,280
is

1161
00:42:45,119 --> 00:42:49,119
the vertical angle so if you're looking

1162
00:42:47,280 --> 00:42:52,400
at something from the side something

1163
00:42:49,119 --> 00:42:53,920
horizontal as you get closer to 90

1164
00:42:52,400 --> 00:42:56,480
degrees you don't see it at all it

1165
00:42:53,920 --> 00:42:58,880
becomes invisible it goes to zero much

1166
00:42:56,480 --> 00:43:01,440
like how cos of any angle goes to zero

1167
00:42:58,880 --> 00:43:04,800
at ninety degrees and similarly for your

1168
00:43:01,440 --> 00:43:06,160
real z times sine of alpha

1169
00:43:04,800 --> 00:43:07,839
it's much the same as you look from

1170
00:43:06,160 --> 00:43:10,800
directly above

1171
00:43:07,839 --> 00:43:14,400
well as you approach zero

1172
00:43:10,800 --> 00:43:17,040
goes to zero which is why you have sign

1173
00:43:14,400 --> 00:43:18,720
so thank you very much all for listening

1174
00:43:17,040 --> 00:43:20,480
i'll be in the comments if you have any

1175
00:43:18,720 --> 00:43:22,319
questions and thank you all to my family

1176
00:43:20,480 --> 00:43:24,839
my amazing teachers miss hogan and mr

1177
00:43:22,319 --> 00:43:27,040
till and i'll see you all

1178
00:43:24,839 --> 00:43:29,040
soon thank you

1179
00:43:27,040 --> 00:43:31,040
very much for listening and have a great

1180
00:43:29,040 --> 00:43:32,960
python

1181
00:43:31,040 --> 00:43:35,280
see you around

1182
00:43:32,960 --> 00:43:38,079
fantastic thanks so much

1183
00:43:35,280 --> 00:43:40,000
now bella is our next speaker and she is

1184
00:43:38,079 --> 00:43:42,240
a repeat offender on the student

1185
00:43:40,000 --> 00:43:44,720
showcase lineup this year we'll be

1186
00:43:42,240 --> 00:43:46,800
hearing about tweet aware using ai to

1187
00:43:44,720 --> 00:43:48,400
deconstruct your twitter timeline

1188
00:43:46,800 --> 00:43:50,240
when i asked bella how long she'd been

1189
00:43:48,400 --> 00:43:53,280
programming she replied nine years in

1190
00:43:50,240 --> 00:43:55,599
total but six years properly

1191
00:43:53,280 --> 00:43:57,920
i'm not sure what designates proper

1192
00:43:55,599 --> 00:44:01,200
programming i assume from improper

1193
00:43:57,920 --> 00:44:02,640
programming but maybe i'll learn one day

1194
00:44:01,200 --> 00:44:05,520
when i asked bella what her favorite

1195
00:44:02,640 --> 00:44:07,920
emoji is she said heart's face which is

1196
00:44:05,520 --> 00:44:09,520
definitely something i can get behind

1197
00:44:07,920 --> 00:44:12,480
when i asked what she's learned recently

1198
00:44:09,520 --> 00:44:15,040
about computing she replied that almost

1199
00:44:12,480 --> 00:44:18,240
90 percent of the world's currency only

1200
00:44:15,040 --> 00:44:20,640
exists on computers so our physical cash

1201
00:44:18,240 --> 00:44:22,560
reserves are only 10

1202
00:44:20,640 --> 00:44:24,560
that's really interesting and i'm really

1203
00:44:22,560 --> 00:44:26,720
interested to hear her talk

1204
00:44:24,560 --> 00:44:28,240
hi my name's bella ryan i'm a year 12

1205
00:44:26,720 --> 00:44:29,920
student at roseville college i'm here

1206
00:44:28,240 --> 00:44:31,599
today to talk about my software design

1207
00:44:29,920 --> 00:44:34,240
development major project

1208
00:44:31,599 --> 00:44:37,760
tweet aware which is a python program

1209
00:44:34,240 --> 00:44:39,599
that uses the twitter api to conduct

1210
00:44:37,760 --> 00:44:40,480
sentiment analysis funding users twitter

1211
00:44:39,599 --> 00:44:43,359
feed

1212
00:44:40,480 --> 00:44:46,240
a bit of background to the project the

1213
00:44:43,359 --> 00:44:49,280
whole idea of this project was to find

1214
00:44:46,240 --> 00:44:51,040
a system either manual or automated

1215
00:44:49,280 --> 00:44:53,440
and find a way that you can

1216
00:44:51,040 --> 00:44:55,200
significantly improve the system

1217
00:44:53,440 --> 00:44:57,119
so

1218
00:44:55,200 --> 00:44:58,720
when it comes to twitter it's not that i

1219
00:44:57,119 --> 00:45:00,880
necessarily found a problem with it it's

1220
00:44:58,720 --> 00:45:03,440
more like an opportunity as we

1221
00:45:00,880 --> 00:45:05,520
increasingly use social media

1222
00:45:03,440 --> 00:45:07,440
to connect with people as it becomes

1223
00:45:05,520 --> 00:45:09,599
more integral in our lives in terms of

1224
00:45:07,440 --> 00:45:12,720
education as well reflection and

1225
00:45:09,599 --> 00:45:15,280
analysis of what we are taking on

1226
00:45:12,720 --> 00:45:19,359
using these platforms becomes more and

1227
00:45:15,280 --> 00:45:20,880
more important uh in 2021 there are 192

1228
00:45:19,359 --> 00:45:23,200
million people actively using twitter in

1229
00:45:20,880 --> 00:45:24,800
the world and out of that user base 500

1230
00:45:23,200 --> 00:45:28,160
million tweets are posted every day

1231
00:45:24,800 --> 00:45:29,280
which is approximately 5 787 points per

1232
00:45:28,160 --> 00:45:30,480
second

1233
00:45:29,280 --> 00:45:32,160
so

1234
00:45:30,480 --> 00:45:33,920
the whole idea is that for each one of

1235
00:45:32,160 --> 00:45:36,800
those tweets there's at least one person

1236
00:45:33,920 --> 00:45:37,839
who's taking on the information so how

1237
00:45:36,800 --> 00:45:39,359
can i

1238
00:45:37,839 --> 00:45:40,160
improve twitter

1239
00:45:39,359 --> 00:45:42,640
to

1240
00:45:40,160 --> 00:45:44,079
give you a better idea of all right this

1241
00:45:42,640 --> 00:45:47,119
is what i'm taking on in terms of the

1242
00:45:44,079 --> 00:45:49,359
idea but reflect on whether

1243
00:45:47,119 --> 00:45:51,040
the way that that idea is put across is

1244
00:45:49,359 --> 00:45:53,040
positive negative or neutral so that you

1245
00:45:51,040 --> 00:45:55,200
can further reflect on how that might

1246
00:45:53,040 --> 00:45:58,000
impact

1247
00:45:55,200 --> 00:45:58,960
other aspects of the information

1248
00:45:58,000 --> 00:46:01,119
so

1249
00:45:58,960 --> 00:46:03,920
first off i started just using the web

1250
00:46:01,119 --> 00:46:06,720
documentation between development uh and

1251
00:46:03,920 --> 00:46:08,240
i found a get request and it was a get

1252
00:46:06,720 --> 00:46:09,760
request for user's timeline i was super

1253
00:46:08,240 --> 00:46:11,680
keen i was like yes this is exactly what

1254
00:46:09,760 --> 00:46:14,640
i'm looking for ran it through with all

1255
00:46:11,680 --> 00:46:16,640
my authentication all of that and i got

1256
00:46:14,640 --> 00:46:18,480
this i was like sweet tweet text it's

1257
00:46:16,640 --> 00:46:20,480
working this is amazing

1258
00:46:18,480 --> 00:46:21,839
uh then when i went across to the

1259
00:46:20,480 --> 00:46:23,920
account that i used it on which was

1260
00:46:21,839 --> 00:46:25,599
twitterdev i realized that what i was

1261
00:46:23,920 --> 00:46:27,119
actually getting was

1262
00:46:25,599 --> 00:46:30,480
all of their tweets read tweets and

1263
00:46:27,119 --> 00:46:33,200
comments so twitter the twitter api

1264
00:46:30,480 --> 00:46:34,960
doesn't actually support you

1265
00:46:33,200 --> 00:46:37,280
just putting in someone's handle and

1266
00:46:34,960 --> 00:46:39,599
then being able to get the feed as they

1267
00:46:37,280 --> 00:46:42,400
see it so i had to recreate

1268
00:46:39,599 --> 00:46:44,480
it and instead model the fade

1269
00:46:42,400 --> 00:46:46,480
uh this is one i used onto postman which

1270
00:46:44,480 --> 00:46:48,160
is an absolutely fabulous

1271
00:46:46,480 --> 00:46:49,920
program to use if you're trying to

1272
00:46:48,160 --> 00:46:52,000
figure out how to use an api especially

1273
00:46:49,920 --> 00:46:54,000
if the people who developed the api have

1274
00:46:52,000 --> 00:46:57,040
created a dashboard like this

1275
00:46:54,000 --> 00:46:59,599
um using that i was able to figure out

1276
00:46:57,040 --> 00:47:01,599
the order of apis i'd need to use to

1277
00:46:59,599 --> 00:47:03,119
recreate someone's feed

1278
00:47:01,599 --> 00:47:04,640
so to do this

1279
00:47:03,119 --> 00:47:07,040
i take on

1280
00:47:04,640 --> 00:47:08,880
someone's i take in input from the user

1281
00:47:07,040 --> 00:47:11,280
which is the handle which i then send to

1282
00:47:08,880 --> 00:47:13,040
the api and i get back their user id

1283
00:47:11,280 --> 00:47:15,119
which i then send to the api again to

1284
00:47:13,040 --> 00:47:16,000
get a list of every um user that they

1285
00:47:15,119 --> 00:47:16,960
follow

1286
00:47:16,000 --> 00:47:19,280
i

1287
00:47:16,960 --> 00:47:21,200
send a request for every single user on

1288
00:47:19,280 --> 00:47:23,680
that list to get all their ids and then

1289
00:47:21,200 --> 00:47:24,720
i use the timeline request that i showed

1290
00:47:23,680 --> 00:47:25,520
earlier

1291
00:47:24,720 --> 00:47:27,200
to

1292
00:47:25,520 --> 00:47:29,040
get their

1293
00:47:27,200 --> 00:47:31,760
um five most recent tweets from their

1294
00:47:29,040 --> 00:47:34,240
account obviously that poses floor

1295
00:47:31,760 --> 00:47:36,000
floors with the model of how i'm

1296
00:47:34,240 --> 00:47:37,359
recreating people's feeds because if

1297
00:47:36,000 --> 00:47:39,200
you've got someone who does two tweets

1298
00:47:37,359 --> 00:47:41,599
in their lifetime versus a journalist

1299
00:47:39,200 --> 00:47:42,640
who does two tweets every five minutes

1300
00:47:41,599 --> 00:47:43,680
uh

1301
00:47:42,640 --> 00:47:46,400
their

1302
00:47:43,680 --> 00:47:49,040
representation in the data

1303
00:47:46,400 --> 00:47:51,359
is not is skewed in a way that's not

1304
00:47:49,040 --> 00:47:55,599
represented representative of an actual

1305
00:47:51,359 --> 00:47:57,119
feed which is why it's a model um i then

1306
00:47:55,599 --> 00:47:58,240
send it through

1307
00:47:57,119 --> 00:48:00,400
a

1308
00:47:58,240 --> 00:48:03,520
scrubbing program that takes out things

1309
00:48:00,400 --> 00:48:05,520
like new lines and other html css

1310
00:48:03,520 --> 00:48:08,319
elements that sometimes comes through in

1311
00:48:05,520 --> 00:48:10,480
the tweet text that are fantastic for

1312
00:48:08,319 --> 00:48:12,720
displaying it on twitter but not very

1313
00:48:10,480 --> 00:48:15,520
useful for the purpose

1314
00:48:12,720 --> 00:48:17,119
of sentiment analysis

1315
00:48:15,520 --> 00:48:18,960
next i send it through the sentiment

1316
00:48:17,119 --> 00:48:21,200
analysis algorithm

1317
00:48:18,960 --> 00:48:25,280
for this i use text blob

1318
00:48:21,200 --> 00:48:26,640
and so i have a list of all of the

1319
00:48:25,280 --> 00:48:29,280
clean tweets

1320
00:48:26,640 --> 00:48:31,920
and i send each tweet through

1321
00:48:29,280 --> 00:48:35,599
text blog which then i use the sentiment

1322
00:48:31,920 --> 00:48:37,440
polarity function to

1323
00:48:35,599 --> 00:48:38,960
give them h a value of either positive

1324
00:48:37,440 --> 00:48:41,040
neutral or negative

1325
00:48:38,960 --> 00:48:42,640
and then when they're outputted from

1326
00:48:41,040 --> 00:48:44,800
that function

1327
00:48:42,640 --> 00:48:47,119
uh they're all input into a dictionary

1328
00:48:44,800 --> 00:48:49,440
where the key is the tweet itself and

1329
00:48:47,119 --> 00:48:50,880
then the value is the

1330
00:48:49,440 --> 00:48:53,440
sentiment of

1331
00:48:50,880 --> 00:48:56,480
mutual and negative

1332
00:48:53,440 --> 00:48:58,880
next is output so i have three kinds of

1333
00:48:56,480 --> 00:49:03,520
output for this project i had a

1334
00:48:58,880 --> 00:49:05,359
txt file that had the percentages

1335
00:49:03,520 --> 00:49:08,160
of positive negative and neutral within

1336
00:49:05,359 --> 00:49:11,119
the data set i had a

1337
00:49:08,160 --> 00:49:12,960
png which is the pie graph that you see

1338
00:49:11,119 --> 00:49:14,160
there of again positive negative and

1339
00:49:12,960 --> 00:49:17,910
neutral

1340
00:49:14,160 --> 00:49:19,280
and then i also had output within the

1341
00:49:17,910 --> 00:49:22,160
[Music]

1342
00:49:19,280 --> 00:49:23,280
command line interface which was the

1343
00:49:22,160 --> 00:49:24,960
five

1344
00:49:23,280 --> 00:49:26,559
the first five in each list of positive

1345
00:49:24,960 --> 00:49:29,920
and negative tweets

1346
00:49:26,559 --> 00:49:33,040
um for the user to look at each time

1347
00:49:29,920 --> 00:49:37,200
they run the program for each handle

1348
00:49:33,040 --> 00:49:38,960
uh from this you i decided to

1349
00:49:37,200 --> 00:49:39,839
run again with a real world example of

1350
00:49:38,960 --> 00:49:42,480
how

1351
00:49:39,839 --> 00:49:45,280
this program could be taken a little bit

1352
00:49:42,480 --> 00:49:47,599
further so as we're in lockdown now a

1353
00:49:45,280 --> 00:49:49,680
lot of us are receiving news through

1354
00:49:47,599 --> 00:49:51,520
social media whether

1355
00:49:49,680 --> 00:49:54,160
it's the um

1356
00:49:51,520 --> 00:49:55,920
11am live stream on facebook

1357
00:49:54,160 --> 00:49:57,520
or it's the little

1358
00:49:55,920 --> 00:49:59,200
uh little bits that have had through

1359
00:49:57,520 --> 00:50:02,720
instagram or twitter in forms of

1360
00:49:59,200 --> 00:50:05,119
infographics and so on so using my

1361
00:50:02,720 --> 00:50:07,920
testing twitter account i followed a

1362
00:50:05,119 --> 00:50:12,400
bunch of australian

1363
00:50:07,920 --> 00:50:15,280
news platforms and then ran them through

1364
00:50:12,400 --> 00:50:17,440
and got the outfit of five most recent

1365
00:50:15,280 --> 00:50:20,160
positive tweets negative tweets tweet

1366
00:50:17,440 --> 00:50:21,920
percentages and then the paragraph

1367
00:50:20,160 --> 00:50:22,720
um

1368
00:50:21,920 --> 00:50:24,559
which

1369
00:50:22,720 --> 00:50:26,079
i think reflects

1370
00:50:24,559 --> 00:50:27,280
the next step where i want this project

1371
00:50:26,079 --> 00:50:29,040
to go

1372
00:50:27,280 --> 00:50:31,280
which is

1373
00:50:29,040 --> 00:50:31,280
uh

1374
00:50:31,520 --> 00:50:34,960
more

1375
00:50:32,400 --> 00:50:38,000
more kind of data visualization so word

1376
00:50:34,960 --> 00:50:40,559
clouds um

1377
00:50:38,000 --> 00:50:43,440
and things like that looking at data

1378
00:50:40,559 --> 00:50:45,359
historically so month to month

1379
00:50:43,440 --> 00:50:47,839
what sentiment scores you're getting

1380
00:50:45,359 --> 00:50:50,079
through your feed and also going back to

1381
00:50:47,839 --> 00:50:51,599
the original plan for my project which

1382
00:50:50,079 --> 00:50:53,839
was to have

1383
00:50:51,599 --> 00:50:56,400
live sentiment analysis on the feed

1384
00:50:53,839 --> 00:50:58,160
itself i think i could achieve that that

1385
00:50:56,400 --> 00:50:59,040
would be really awesome

1386
00:50:58,160 --> 00:51:00,960
uh

1387
00:50:59,040 --> 00:51:02,800
yeah so thank you for listening to my

1388
00:51:00,960 --> 00:51:05,040
presentation i guess the main takeaway

1389
00:51:02,800 --> 00:51:07,119
that i want from this is when you're

1390
00:51:05,040 --> 00:51:08,800
given the chance to do a project in

1391
00:51:07,119 --> 00:51:10,000
school pick something that you love and

1392
00:51:08,800 --> 00:51:11,839
pick something that you can keep on

1393
00:51:10,000 --> 00:51:13,680
working on because

1394
00:51:11,839 --> 00:51:16,240
during the hard knock times of the

1395
00:51:13,680 --> 00:51:18,319
project itself it kind of prompts you to

1396
00:51:16,240 --> 00:51:20,079
actually continue working on the project

1397
00:51:18,319 --> 00:51:21,760
but afterwards it gives you a fantastic

1398
00:51:20,079 --> 00:51:22,800
springboard to jump off into your next

1399
00:51:21,760 --> 00:51:24,480
project

1400
00:51:22,800 --> 00:51:26,880
thank you so much

1401
00:51:24,480 --> 00:51:28,960
now our final speaker today is menindra

1402
00:51:26,880 --> 00:51:31,200
who'll be talking about de-blurring with

1403
00:51:28,960 --> 00:51:32,319
deep learning which is perhaps more of a

1404
00:51:31,200 --> 00:51:34,160
tongue twister than i should have

1405
00:51:32,319 --> 00:51:36,079
finished today on

1406
00:51:34,160 --> 00:51:38,880
meninja has been programming for three

1407
00:51:36,079 --> 00:51:41,119
years and his favorite emoji is the dead

1408
00:51:38,880 --> 00:51:43,760
face emoji specifically he said he

1409
00:51:41,119 --> 00:51:45,680
thinks it's quite funny

1410
00:51:43,760 --> 00:51:47,040
when i asked what recent topic of

1411
00:51:45,680 --> 00:51:49,520
computer science he's been learning

1412
00:51:47,040 --> 00:51:51,440
about he said machine learning deep

1413
00:51:49,520 --> 00:51:53,680
learning and specifically supervised

1414
00:51:51,440 --> 00:51:57,040
learning with neural networks

1415
00:51:53,680 --> 00:51:59,760
i'm really excited to hear his talk

1416
00:51:57,040 --> 00:52:01,440
hello pycon my name is melinda de mel

1417
00:51:59,760 --> 00:52:04,480
and i'd like to present my project d

1418
00:52:01,440 --> 00:52:06,079
blurring with deep learning so i first

1419
00:52:04,480 --> 00:52:08,079
began the year writing a report on back

1420
00:52:06,079 --> 00:52:09,839
propagation and the report presents an

1421
00:52:08,079 --> 00:52:11,599
intuitive and comprehensive analysis on

1422
00:52:09,839 --> 00:52:13,280
the algorithm it was a lot of fun to

1423
00:52:11,599 --> 00:52:14,720
write

1424
00:52:13,280 --> 00:52:15,839
so i then wanted to extend my newly

1425
00:52:14,720 --> 00:52:18,640
found knowledge under a couple of

1426
00:52:15,839 --> 00:52:20,800
project ideas so i thought of q learning

1427
00:52:18,640 --> 00:52:22,640
the genetic algorithm neat machine

1428
00:52:20,800 --> 00:52:23,760
learning with robotics quantum machine

1429
00:52:22,640 --> 00:52:25,760
learning

1430
00:52:23,760 --> 00:52:27,599
i thought of fluid simulation binary

1431
00:52:25,760 --> 00:52:29,359
systems

1432
00:52:27,599 --> 00:52:31,359
but then i realized that all of these

1433
00:52:29,359 --> 00:52:33,280
ideas were kind of irrelevant to back

1434
00:52:31,359 --> 00:52:34,480
propagation and out of scope for my

1435
00:52:33,280 --> 00:52:36,720
project

1436
00:52:34,480 --> 00:52:38,240
so i then moved on to computer vision

1437
00:52:36,720 --> 00:52:39,599
and of computer vision i chose image

1438
00:52:38,240 --> 00:52:41,040
restoration

1439
00:52:39,599 --> 00:52:42,880
i chose it mainly because it was

1440
00:52:41,040 --> 00:52:44,160
relevant to back propagation and

1441
00:52:42,880 --> 00:52:45,839
actually initially i was going to

1442
00:52:44,160 --> 00:52:47,920
restore color into black and white

1443
00:52:45,839 --> 00:52:49,839
images but then i realized this is kind

1444
00:52:47,920 --> 00:52:52,160
of deprecated because no one takes black

1445
00:52:49,839 --> 00:52:53,760
on my photos these days so then uh i

1446
00:52:52,160 --> 00:52:55,280
moved on to d blurring and i realized

1447
00:52:53,760 --> 00:52:57,760
it's more relevant because people do

1448
00:52:55,280 --> 00:52:59,599
take blurred photos these days and there

1449
00:52:57,760 --> 00:53:02,319
aren't many implementations of solutions

1450
00:52:59,599 --> 00:53:04,160
out there to resolve this

1451
00:53:02,319 --> 00:53:06,720
i'll now go through a brief overview of

1452
00:53:04,160 --> 00:53:08,559
how deep learning works so first we have

1453
00:53:06,720 --> 00:53:10,640
neural networks and neural networks are

1454
00:53:08,559 --> 00:53:12,079
modeled off of the human brain where we

1455
00:53:10,640 --> 00:53:14,000
can think of our brain as a bunch of

1456
00:53:12,079 --> 00:53:15,839
neurons connected to other neurons

1457
00:53:14,000 --> 00:53:18,000
through axioms

1458
00:53:15,839 --> 00:53:19,920
and similarly in neural networks we have

1459
00:53:18,000 --> 00:53:21,440
layers of neurons

1460
00:53:19,920 --> 00:53:22,240
that connect to other layers of neurons

1461
00:53:21,440 --> 00:53:23,920
with

1462
00:53:22,240 --> 00:53:26,079
weights where the blue ones are positive

1463
00:53:23,920 --> 00:53:29,359
and the red ones are negative and they

1464
00:53:26,079 --> 00:53:31,599
all affect each other accordingly

1465
00:53:29,359 --> 00:53:33,839
we also have biases which are associated

1466
00:53:31,599 --> 00:53:36,000
with each neuron and they just help with

1467
00:53:33,839 --> 00:53:37,520
the training process

1468
00:53:36,000 --> 00:53:39,760
so here's what a neuron might look like

1469
00:53:37,520 --> 00:53:43,119
we have the weights the bias the

1470
00:53:39,760 --> 00:53:44,480
activation and the way to calculate it

1471
00:53:43,119 --> 00:53:46,480
and here's what the network might look

1472
00:53:44,480 --> 00:53:48,880
like we have the layers

1473
00:53:46,480 --> 00:53:51,359
and the layer sizes and we also have a

1474
00:53:48,880 --> 00:53:55,880
way to generate the initial uh

1475
00:53:51,359 --> 00:53:55,880
weights connecting layers together

1476
00:53:56,079 --> 00:54:00,079
propagation uh

1477
00:53:57,520 --> 00:54:03,200
for progression simply put it's just

1478
00:54:00,079 --> 00:54:05,520
taking an input and then applying uh the

1479
00:54:03,200 --> 00:54:07,920
weights and biases from each layer and

1480
00:54:05,520 --> 00:54:10,800
then passing it to the network uh in

1481
00:54:07,920 --> 00:54:13,040
real life this could be compared to um

1482
00:54:10,800 --> 00:54:15,359
taking an input from our environment our

1483
00:54:13,040 --> 00:54:18,160
brain processing it and giving us an

1484
00:54:15,359 --> 00:54:20,000
up an output

1485
00:54:18,160 --> 00:54:21,760
so initially for a neural network this

1486
00:54:20,000 --> 00:54:24,240
causes high error between the expected

1487
00:54:21,760 --> 00:54:25,920
output and the network's output but once

1488
00:54:24,240 --> 00:54:28,880
it's been trained

1489
00:54:25,920 --> 00:54:30,640
it's actually what's used to predict uh

1490
00:54:28,880 --> 00:54:33,040
non-blood images in my case or the

1491
00:54:30,640 --> 00:54:33,040
answer

1492
00:54:33,920 --> 00:54:37,520
for propagation

1493
00:54:35,680 --> 00:54:39,440
this is what it would look like in code

1494
00:54:37,520 --> 00:54:40,960
so we have

1495
00:54:39,440 --> 00:54:42,319
we go through each layer and we go

1496
00:54:40,960 --> 00:54:44,559
through each node in the layer and we

1497
00:54:42,319 --> 00:54:46,319
pass it the previous layer

1498
00:54:44,559 --> 00:54:48,640
and then we

1499
00:54:46,319 --> 00:54:50,000
calculate all the weights and all the

1500
00:54:48,640 --> 00:54:52,000
activations from the previous layer and

1501
00:54:50,000 --> 00:54:55,280
we find an activation for that node in

1502
00:54:52,000 --> 00:54:55,280
the next layer

1503
00:54:55,599 --> 00:55:00,079
so this is what the network might look

1504
00:54:56,799 --> 00:55:01,440
like so we could uh take the input we

1505
00:55:00,079 --> 00:55:03,040
pass it to

1506
00:55:01,440 --> 00:55:05,040
the processing layer or the hidden layer

1507
00:55:03,040 --> 00:55:06,240
and get an activation and then from this

1508
00:55:05,040 --> 00:55:08,400
we pass through the next layer the

1509
00:55:06,240 --> 00:55:10,960
output and we get activations for all

1510
00:55:08,400 --> 00:55:12,960
these accordingly

1511
00:55:10,960 --> 00:55:15,040
uh back propagation so back propagation

1512
00:55:12,960 --> 00:55:18,240
really is the focus of my algorithm it's

1513
00:55:15,040 --> 00:55:20,480
like the uh heart of the algorithm where

1514
00:55:18,240 --> 00:55:23,119
without it it won't be able to function

1515
00:55:20,480 --> 00:55:25,040
uh the structure of uh

1516
00:55:23,119 --> 00:55:26,400
the neural network itself uh would

1517
00:55:25,040 --> 00:55:28,000
effectively be useless without back

1518
00:55:26,400 --> 00:55:29,440
propagation because you won't be able to

1519
00:55:28,000 --> 00:55:30,319
learn

1520
00:55:29,440 --> 00:55:32,240
so

1521
00:55:30,319 --> 00:55:34,079
here's the mathematics for it so

1522
00:55:32,240 --> 00:55:35,520
i don't want to read any of it but this

1523
00:55:34,079 --> 00:55:37,599
is for my report and i just want you to

1524
00:55:35,520 --> 00:55:40,160
acknowledge it it was a lot of fun to uh

1525
00:55:37,599 --> 00:55:41,599
learn about read about and write about

1526
00:55:40,160 --> 00:55:42,720
and essentially what we're going to take

1527
00:55:41,599 --> 00:55:44,720
is this

1528
00:55:42,720 --> 00:55:45,760
uh derivative or

1529
00:55:44,720 --> 00:55:46,799
um

1530
00:55:45,760 --> 00:55:49,200
equation

1531
00:55:46,799 --> 00:55:51,920
per se so you want to find the rate of

1532
00:55:49,200 --> 00:55:54,799
change of the cost with respect to each

1533
00:55:51,920 --> 00:55:57,599
weight and bias of the network so the

1534
00:55:54,799 --> 00:55:58,640
cost is the overall error of the network

1535
00:55:57,599 --> 00:56:00,960
it's the

1536
00:55:58,640 --> 00:56:02,960
network's output and the expected output

1537
00:56:00,960 --> 00:56:05,760
and the difference and basically we want

1538
00:56:02,960 --> 00:56:07,680
to minimize this to make the network get

1539
00:56:05,760 --> 00:56:09,359
the expected output

1540
00:56:07,680 --> 00:56:11,599
and so effectively what do we apply the

1541
00:56:09,359 --> 00:56:13,440
chain rule to each meter as major aspect

1542
00:56:11,599 --> 00:56:14,720
of the

1543
00:56:13,440 --> 00:56:17,680
neural network

1544
00:56:14,720 --> 00:56:20,000
and then we uh

1545
00:56:17,680 --> 00:56:22,400
find how each rate and bias

1546
00:56:20,000 --> 00:56:24,079
affect the overall cost and we can see

1547
00:56:22,400 --> 00:56:26,319
these how this is written in the code so

1548
00:56:24,079 --> 00:56:27,280
we have these highlighted boxes and you

1549
00:56:26,319 --> 00:56:29,040
can see

1550
00:56:27,280 --> 00:56:30,319
the yellow blocks we take the overall

1551
00:56:29,040 --> 00:56:33,280
cost respect to the output of the

1552
00:56:30,319 --> 00:56:35,359
network in the blue box we take the

1553
00:56:33,280 --> 00:56:36,960
output layer with respect to all the

1554
00:56:35,359 --> 00:56:39,680
previous layers

1555
00:56:36,960 --> 00:56:42,160
and then for each neuron in the previous

1556
00:56:39,680 --> 00:56:44,559
layers we take the neuron with respect

1557
00:56:42,160 --> 00:56:46,720
to each made and bias and we get the

1558
00:56:44,559 --> 00:56:49,040
each uh each weight and biases error on

1559
00:56:46,720 --> 00:56:50,240
the overall cost

1560
00:56:49,040 --> 00:56:52,960
and so this is what it might look like

1561
00:56:50,240 --> 00:56:55,359
so we have uh the input and then these

1562
00:56:52,960 --> 00:56:57,440
numbers are the effect on the overall

1563
00:56:55,359 --> 00:56:59,599
cost so this neuron here has a negative

1564
00:56:57,440 --> 00:57:01,040
effect on the overall cost and then this

1565
00:56:59,599 --> 00:57:02,799
neuron has a positive effect on your

1566
00:57:01,040 --> 00:57:04,960
overall cost and basically you want to

1567
00:57:02,799 --> 00:57:08,400
make this as close as possible to zero

1568
00:57:04,960 --> 00:57:10,400
so we get the expected output

1569
00:57:08,400 --> 00:57:13,680
and so once we have these errors and

1570
00:57:10,400 --> 00:57:15,119
these costs for each neuron we can

1571
00:57:13,680 --> 00:57:16,720
affect the

1572
00:57:15,119 --> 00:57:19,280
rates and biases accordingly by

1573
00:57:16,720 --> 00:57:22,000
increasing them up or down to reduce

1574
00:57:19,280 --> 00:57:22,000
this cost

1575
00:57:22,160 --> 00:57:27,440
and so once the network has been trained

1576
00:57:23,839 --> 00:57:29,760
we can then um save these the the

1577
00:57:27,440 --> 00:57:32,000
parameters or the weights and biases and

1578
00:57:29,760 --> 00:57:34,400
then we can use for propagation to

1579
00:57:32,000 --> 00:57:35,760
simply pass an input and get a predicted

1580
00:57:34,400 --> 00:57:37,119
output which hopefully is really

1581
00:57:35,760 --> 00:57:39,119
accurate

1582
00:57:37,119 --> 00:57:40,720
so here's some training data so here

1583
00:57:39,119 --> 00:57:43,760
here we have the error or the cost on

1584
00:57:40,720 --> 00:57:46,400
the y-axis and we have the iterations uh

1585
00:57:43,760 --> 00:57:47,760
the training process and we can see we

1586
00:57:46,400 --> 00:57:49,839
want to minimize this cost and it's

1587
00:57:47,760 --> 00:57:50,960
doing this but it's still relatively

1588
00:57:49,839 --> 00:57:53,520
high

1589
00:57:50,960 --> 00:57:55,760
and you can see in this one as well

1590
00:57:53,520 --> 00:57:58,720
it goes down but it's still like above

1591
00:57:55,760 --> 00:58:01,200
20 and this is quite bad and you can see

1592
00:57:58,720 --> 00:58:03,839
the result of this as well so

1593
00:58:01,200 --> 00:58:06,400
we have our blurred image we have our

1594
00:58:03,839 --> 00:58:09,280
sharpen image and then we have

1595
00:58:06,400 --> 00:58:11,440
an output so this is a complete mess

1596
00:58:09,280 --> 00:58:12,559
it's completely unrecognizable from the

1597
00:58:11,440 --> 00:58:15,200
other two

1598
00:58:12,559 --> 00:58:16,559
and so i figured that maybe these images

1599
00:58:15,200 --> 00:58:18,880
were too big

1600
00:58:16,559 --> 00:58:20,720
their dimensions were like 500 by 637

1601
00:58:18,880 --> 00:58:23,280
pixels which is a lot of way to biases

1602
00:58:20,720 --> 00:58:26,480
to compute so i moved to a smaller data

1603
00:58:23,280 --> 00:58:28,720
set and i use the mnist the blurring set

1604
00:58:26,480 --> 00:58:31,119
for digits handwritten digits and we can

1605
00:58:28,720 --> 00:58:33,599
see a blurred six a sharpen six

1606
00:58:31,119 --> 00:58:35,280
and the network's output which again is

1607
00:58:33,599 --> 00:58:36,640
still pretty bad but you can see that

1608
00:58:35,280 --> 00:58:39,440
like the bottom right half is being

1609
00:58:36,640 --> 00:58:41,280
sharpened which is kind of good

1610
00:58:39,440 --> 00:58:43,359
and then i realized that maybe it's due

1611
00:58:41,280 --> 00:58:46,160
to the training method so i changed from

1612
00:58:43,359 --> 00:58:48,240
stochastic going by each row

1613
00:58:46,160 --> 00:58:50,480
uh row by row sorry and then i just

1614
00:58:48,240 --> 00:58:53,119
posted the whole image to the network

1615
00:58:50,480 --> 00:58:56,000
and we saw that the

1616
00:58:53,119 --> 00:58:58,799
loss or the cost on the error decreased

1617
00:58:56,000 --> 00:59:01,280
very fast it went from like 200 to like

1618
00:58:58,799 --> 00:59:02,559
below 10 in like the first iteration and

1619
00:59:01,280 --> 00:59:05,680
it stayed there which is really really

1620
00:59:02,559 --> 00:59:07,280
good and we can see we get some results

1621
00:59:05,680 --> 00:59:09,839
from this so we have a blurred five and

1622
00:59:07,280 --> 00:59:11,839
a sharpen five a couple of results i got

1623
00:59:09,839 --> 00:59:13,359
and these are very sharp you can see the

1624
00:59:11,839 --> 00:59:14,880
network really picked up on the edges

1625
00:59:13,359 --> 00:59:16,160
and the curves of the fives

1626
00:59:14,880 --> 00:59:20,240
you can see some more examples of

1627
00:59:16,160 --> 00:59:22,400
blurred five we have these um

1628
00:59:20,240 --> 00:59:24,079
some more fives these are

1629
00:59:22,400 --> 00:59:26,000
some of the better ones i had especially

1630
00:59:24,079 --> 00:59:28,000
this one really got the shape of the fly

1631
00:59:26,000 --> 00:59:30,720
if not if it's actually better than the

1632
00:59:28,000 --> 00:59:33,760
answer in my opinion

1633
00:59:30,720 --> 00:59:36,000
and so now i have two demos to show my

1634
00:59:33,760 --> 00:59:38,079
uh project so the first demo is the

1635
00:59:36,000 --> 00:59:39,680
training process

1636
00:59:38,079 --> 00:59:42,079
so we have a video here

1637
00:59:39,680 --> 00:59:44,079
so we have the blurred seven we have the

1638
00:59:42,079 --> 00:59:45,680
sharpen seven and then we have the

1639
00:59:44,079 --> 00:59:47,680
network output down here and we can have

1640
00:59:45,680 --> 00:59:49,520
the graph showing the error with respect

1641
00:59:47,680 --> 00:59:52,000
to each training iteration and we can

1642
00:59:49,520 --> 00:59:54,720
see as it goes it slowly

1643
00:59:52,000 --> 00:59:57,440
the error slowly decreases and slowly

1644
00:59:54,720 --> 00:59:59,359
the server uh the network finds

1645
00:59:57,440 --> 01:00:01,280
a really good seven after a couple

1646
00:59:59,359 --> 01:00:04,079
iterations

1647
01:00:01,280 --> 01:00:06,160
like that and you can see again blurred

1648
01:00:04,079 --> 01:00:07,680
sharpen and a really really good output

1649
01:00:06,160 --> 01:00:09,599
right here

1650
01:00:07,680 --> 01:00:12,000
and my second

1651
01:00:09,599 --> 01:00:13,920
demo is going to be a network output so

1652
01:00:12,000 --> 01:00:15,599
we can see we have a blurred floor here

1653
01:00:13,920 --> 01:00:17,440
a sharpened floor here and if i just

1654
01:00:15,599 --> 01:00:20,240
fall propagate through a network

1655
01:00:17,440 --> 01:00:22,240
and pass it this blurred four

1656
01:00:20,240 --> 01:00:24,720
just like so

1657
01:00:22,240 --> 01:00:25,920
and then pass it right here

1658
01:00:24,720 --> 01:00:26,880
you can see

1659
01:00:25,920 --> 01:00:29,200
boom

1660
01:00:26,880 --> 01:00:31,280
blood fall sharpen floor and then our

1661
01:00:29,200 --> 01:00:33,280
network output which again is really

1662
01:00:31,280 --> 01:00:34,559
sharp like it's almost it's actually

1663
01:00:33,280 --> 01:00:37,359
better than the

1664
01:00:34,559 --> 01:00:39,680
actual answer

1665
01:00:37,359 --> 01:00:42,400
and so now

1666
01:00:39,680 --> 01:00:46,000
um

1667
01:00:42,400 --> 01:00:47,920
we have that's our pro my project so

1668
01:00:46,000 --> 01:00:50,240
overall this project was a lot of fun to

1669
01:00:47,920 --> 01:00:51,599
write uh write the code for it was a lot

1670
01:00:50,240 --> 01:00:54,640
of fun to learn about it was a good

1671
01:00:51,599 --> 01:00:55,920
experience i had a lot of fun

1672
01:00:54,640 --> 01:00:57,040
and i recommend it to anyone out there

1673
01:00:55,920 --> 01:00:58,480
who wants to start machine learning

1674
01:00:57,040 --> 01:01:00,720
because this is my machine first machine

1675
01:00:58,480 --> 01:01:02,240
learning project and if anyone has any

1676
01:01:00,720 --> 01:01:04,400
questions about a project or how it

1677
01:01:02,240 --> 01:01:07,280
works just leave it in the chat and

1678
01:01:04,400 --> 01:01:09,440
thank you for watching my presentation

1679
01:01:07,280 --> 01:01:11,599
thank you so much for joining me and

1680
01:01:09,440 --> 01:01:13,599
thank you again to our fantastic

1681
01:01:11,599 --> 01:01:15,359
speakers and i will pass it back to

1682
01:01:13,599 --> 01:01:17,119
future nikki

1683
01:01:15,359 --> 01:01:18,720
to quote one of my

1684
01:01:17,119 --> 01:01:20,640
favorite books by josh barrows on

1685
01:01:18,720 --> 01:01:23,359
fourier transforms

1686
01:01:20,640 --> 01:01:25,280
don't ask really don't ask unless you're

1687
01:01:23,359 --> 01:01:27,119
willing to put in a couple years of

1688
01:01:25,280 --> 01:01:28,319
degree level maths i still have

1689
01:01:27,119 --> 01:01:31,200
nightmares

1690
01:01:28,319 --> 01:01:33,760
to be honest i still don't understand it

1691
01:01:31,200 --> 01:01:35,599
after about two months i finally got it

1692
01:01:33,760 --> 01:01:37,440
the microphone was picking up sound and

1693
01:01:35,599 --> 01:01:39,839
calculating the transform of certain

1694
01:01:37,440 --> 01:01:41,440
frequencies i couldn't believe it now

1695
01:01:39,839 --> 01:01:44,240
all i had to do was send it to the

1696
01:01:41,440 --> 01:01:47,520
output that's where i was wrong the pico

1697
01:01:44,240 --> 01:01:50,480
even with a dual core arm cortex mo plus

1698
01:01:47,520 --> 01:01:53,040
processor with 264 kilobytes internal

1699
01:01:50,480 --> 01:01:54,079
ram and support for up to 60 megabytes

1700
01:01:53,040 --> 01:01:56,640
of off

1701
01:01:54,079 --> 01:01:59,839
chip flash it just didn't have the power

1702
01:01:56,640 --> 01:02:02,160
to do proper audio processing

1703
01:01:59,839 --> 01:02:04,240
i could do everything separately but as

1704
01:02:02,160 --> 01:02:07,200
soon as i tried to put it all together

1705
01:02:04,240 --> 01:02:08,559
it just didn't work i rang around in

1706
01:02:07,200 --> 01:02:10,240
hopes that there might be something i

1707
01:02:08,559 --> 01:02:12,480
could do but no

1708
01:02:10,240 --> 01:02:15,359
everyone told me that i had to move on

1709
01:02:12,480 --> 01:02:17,920
to something called potency

1710
01:02:15,359 --> 01:02:19,359
i was a little bit upset

1711
01:02:17,920 --> 01:02:21,760
i had spent months trying to get it to

1712
01:02:19,359 --> 01:02:23,680
work on python and figured out how to do

1713
01:02:21,760 --> 01:02:25,359
a fourier transform only to find that it

1714
01:02:23,680 --> 01:02:27,359
failed at the last step

1715
01:02:25,359 --> 01:02:30,319
but there was no time to be sad

1716
01:02:27,359 --> 01:02:32,480
so i bought a teensy and an audio shield

1717
01:02:30,319 --> 01:02:35,520
and immediately blew them both up but

1718
01:02:32,480 --> 01:02:40,440
there was no time to be sad so i bought

1719
01:02:35,520 --> 01:03:55,999
another tinsy and this time it worked

1720
01:02:40,440 --> 01:03:55,999
[Music]

1721
01:03:59,520 --> 01:04:24,619
[Music]

1722
01:04:26,160 --> 01:04:30,960
using the handy audio library built for

1723
01:04:28,559 --> 01:04:33,280
tinsy and the graphic interface for

1724
01:04:30,960 --> 01:04:35,359
writing code i managed to build

1725
01:04:33,280 --> 01:04:38,079
something resembling a hearing aid the

1726
01:04:35,359 --> 01:04:40,799
code for the fourier transform was so

1727
01:04:38,079 --> 01:04:42,319
easy with the audio design system it was

1728
01:04:40,799 --> 01:04:44,559
my saving grace

1729
01:04:42,319 --> 01:04:46,559
i was able to send the audio via

1730
01:04:44,559 --> 01:04:49,200
bluetooth to a receiver that sits in the

1731
01:04:46,559 --> 01:04:50,799
air using an old hearing aid speaker i

1732
01:04:49,200 --> 01:04:52,640
had lying around

1733
01:04:50,799 --> 01:04:54,720
building a hearing aid from scratch

1734
01:04:52,640 --> 01:04:55,599
wasn't easy but i never expected it to

1735
01:04:54,720 --> 01:04:57,119
be

1736
01:04:55,599 --> 01:04:59,760
growing up with hearing loss i

1737
01:04:57,119 --> 01:05:02,240
understand how important it is to have

1738
01:04:59,760 --> 01:05:03,920
the resources available so that i have

1739
01:05:02,240 --> 01:05:05,920
the same chance as everyone else around

1740
01:05:03,920 --> 01:05:08,240
me to go to school partake in

1741
01:05:05,920 --> 01:05:10,880
discussions about my health and develop

1742
01:05:08,240 --> 01:05:15,440
a sustainable and equal future and that

1743
01:05:10,880 --> 01:05:19,839
is why i built renewable hearing

1744
01:05:15,440 --> 01:05:20,640
[Music]

1745
01:05:19,839 --> 01:05:22,880
okay

1746
01:05:20,640 --> 01:05:24,720
thanks bye

1747
01:05:22,880 --> 01:05:27,359
thank you so much liv

1748
01:05:24,720 --> 01:05:29,599
up next is jonty jonty's talking about

1749
01:05:27,359 --> 01:05:30,960
python meta programming through bytecode

1750
01:05:29,599 --> 01:05:32,319
and has been learning to program for

1751
01:05:30,960 --> 01:05:34,240
three years now

1752
01:05:32,319 --> 01:05:36,559
his favorite emoji is the smiley face

1753
01:05:34,240 --> 01:05:38,720
emoji although technically he sent

1754
01:05:36,559 --> 01:05:41,440
through a colon and a bracket but i

1755
01:05:38,720 --> 01:05:42,799
think i'm being a little pedantic here

1756
01:05:41,440 --> 01:05:44,640
when i asked what he's been recently

1757
01:05:42,799 --> 01:05:46,480
learning about computers he said he's

1758
01:05:44,640 --> 01:05:48,000
been learning about the axioms of

1759
01:05:46,480 --> 01:05:50,880
natural deduction

1760
01:05:48,000 --> 01:05:52,559
let's hear his talk i'm uh john t deakin

1761
01:05:50,880 --> 01:05:53,599
and this is a smart project by code

1762
01:05:52,559 --> 01:05:56,480
macro

1763
01:05:53,599 --> 01:05:58,960
uh it's a look at intermediate languages

1764
01:05:56,480 --> 01:06:00,960
and meta programming uh specifically in

1765
01:05:58,960 --> 01:06:03,039
the context of like fourth and lisp

1766
01:06:00,960 --> 01:06:04,960
macros um

1767
01:06:03,039 --> 01:06:06,319
yeah so this is just a cool example i

1768
01:06:04,960 --> 01:06:07,839
thought i'd flash up at the start we'll

1769
01:06:06,319 --> 01:06:10,160
go through it um

1770
01:06:07,839 --> 01:06:11,599
so i'm gonna go through some of the

1771
01:06:10,160 --> 01:06:12,880
you know intermediate languages and meta

1772
01:06:11,599 --> 01:06:14,960
programming stuff just make sure we're

1773
01:06:12,880 --> 01:06:16,240
all on the same page uh before i start

1774
01:06:14,960 --> 01:06:20,079
talking about

1775
01:06:16,240 --> 01:06:22,400
the bytecode macro library itself um

1776
01:06:20,079 --> 01:06:23,760
so intermediate languages what are they

1777
01:06:22,400 --> 01:06:24,640
uh

1778
01:06:23,760 --> 01:06:26,640
when

1779
01:06:24,640 --> 01:06:28,559
a compiler runs it doesn't just run the

1780
01:06:26,640 --> 01:06:30,079
string directly you know that would be

1781
01:06:28,559 --> 01:06:32,000
insane it goes through a few steps

1782
01:06:30,079 --> 01:06:33,680
beforehand

1783
01:06:32,000 --> 01:06:35,839
and you know in this example here we've

1784
01:06:33,680 --> 01:06:38,640
got is a string going through alexa to

1785
01:06:35,839 --> 01:06:40,720
be uh to become tokens which is then

1786
01:06:38,640 --> 01:06:42,880
passed to become an ast which is then

1787
01:06:40,720 --> 01:06:44,480
compiled to become bytecode and then the

1788
01:06:42,880 --> 01:06:46,880
by code's interpreted

1789
01:06:44,480 --> 01:06:48,400
later and it's that by code that's what

1790
01:06:46,880 --> 01:06:50,079
we're looking at today

1791
01:06:48,400 --> 01:06:52,079
um well there's not always by code this

1792
01:06:50,079 --> 01:06:53,680
is just a overview of python itself it

1793
01:06:52,079 --> 01:06:54,960
does you know it changes from compiler

1794
01:06:53,680 --> 01:06:56,160
to compiler

1795
01:06:54,960 --> 01:06:57,839
uh

1796
01:06:56,160 --> 01:07:00,880
but yeah so in python it's a byte code

1797
01:06:57,839 --> 01:07:03,920
and a bytecode is like a series of bytes

1798
01:07:00,880 --> 01:07:05,440
that is a shortened representation of

1799
01:07:03,920 --> 01:07:07,280
the code and simplified and shortened

1800
01:07:05,440 --> 01:07:08,559
representation of the code you can do

1801
01:07:07,280 --> 01:07:10,799
some cool things with it it's way easier

1802
01:07:08,559 --> 01:07:12,400
to interpret and you can do some cool

1803
01:07:10,799 --> 01:07:14,160
optimizations uh

1804
01:07:12,400 --> 01:07:16,480
not so much with stack base by code

1805
01:07:14,160 --> 01:07:18,799
which python is um and so stack based

1806
01:07:16,480 --> 01:07:20,000
back code is a byte code that primarily

1807
01:07:18,799 --> 01:07:21,520
acts upon

1808
01:07:20,000 --> 01:07:23,520
a stack

1809
01:07:21,520 --> 01:07:25,039
uh a data set you know just all the

1810
01:07:23,520 --> 01:07:26,319
other kinds of things but mostly data

1811
01:07:25,039 --> 01:07:27,119
right the data sets what we're talking

1812
01:07:26,319 --> 01:07:30,000
about

1813
01:07:27,119 --> 01:07:30,960
and so in this case we've got it loading

1814
01:07:30,000 --> 01:07:33,440
on the right side here we've got it

1815
01:07:30,960 --> 01:07:34,559
loading a string hello world onto the

1816
01:07:33,440 --> 01:07:36,400
stack

1817
01:07:34,559 --> 01:07:38,640
and then you can pop that string into

1818
01:07:36,400 --> 01:07:39,680
standard out using the print command

1819
01:07:38,640 --> 01:07:42,000
right

1820
01:07:39,680 --> 01:07:43,520
and so that's you know how python by

1821
01:07:42,000 --> 01:07:45,280
code mostly works just a lot of these

1822
01:07:43,520 --> 01:07:46,319
cool stack operations

1823
01:07:45,280 --> 01:07:47,599
um

1824
01:07:46,319 --> 01:07:48,480
and

1825
01:07:47,599 --> 01:07:50,079
uh

1826
01:07:48,480 --> 01:07:51,520
yeah and so we'll get into meta

1827
01:07:50,079 --> 01:07:54,160
programming now

1828
01:07:51,520 --> 01:07:55,839
meta programming is you know code acting

1829
01:07:54,160 --> 01:07:57,359
upon code it's actually everywhere i i

1830
01:07:55,839 --> 01:07:59,359
didn't realize this before it's checked

1831
01:07:57,359 --> 01:08:00,720
but it's like almost every language is

1832
01:07:59,359 --> 01:08:03,119
made up for some meta programming

1833
01:08:00,720 --> 01:08:04,799
capabilities uh you know java c-sharp

1834
01:08:03,119 --> 01:08:07,280
has reflection

1835
01:08:04,799 --> 01:08:10,240
julie uh python has meta classes c plus

1836
01:08:07,280 --> 01:08:12,000
plus famously template meta programming

1837
01:08:10,240 --> 01:08:14,960
uh and fourth

1838
01:08:12,000 --> 01:08:17,520
fourth uh lisp and julia

1839
01:08:14,960 --> 01:08:20,000
all have this really cool kind of macros

1840
01:08:17,520 --> 01:08:22,080
thing i mean fourth and julia are based

1841
01:08:20,000 --> 01:08:23,839
you know inspired by this but

1842
01:08:22,080 --> 01:08:24,960
they all have a really cool macro system

1843
01:08:23,839 --> 01:08:25,920
where you can

1844
01:08:24,960 --> 01:08:28,000
take

1845
01:08:25,920 --> 01:08:30,640
uh any co you can take some code and

1846
01:08:28,000 --> 01:08:32,000
then transform it into new code

1847
01:08:30,640 --> 01:08:34,640
uh and

1848
01:08:32,000 --> 01:08:36,799
uh you know you can extend the language

1849
01:08:34,640 --> 01:08:38,640
you can write shorthands for some stuff

1850
01:08:36,799 --> 01:08:40,560
you know with code reuse you can even

1851
01:08:38,640 --> 01:08:42,880
like take it so far as to write domain

1852
01:08:40,560 --> 01:08:45,040
specific languages in other language in

1853
01:08:42,880 --> 01:08:47,199
your own language i think it's very cool

1854
01:08:45,040 --> 01:08:50,159
uh so for example like in fixed maths

1855
01:08:47,199 --> 01:08:52,400
you can do infinix maths in lisp

1856
01:08:50,159 --> 01:08:54,239
some guys done that i think it's cool uh

1857
01:08:52,400 --> 01:08:56,239
and for this project i specifically was

1858
01:08:54,239 --> 01:08:57,440
looking mostly at fourth macros just

1859
01:08:56,239 --> 01:08:59,120
because i think

1860
01:08:57,440 --> 01:09:01,199
they've applied better and you know

1861
01:08:59,120 --> 01:09:03,440
that's what i was learning at the time

1862
01:09:01,199 --> 01:09:05,920
uh and the question is why if these are

1863
01:09:03,440 --> 01:09:07,359
so cool and such as you know they're

1864
01:09:05,920 --> 01:09:09,759
very beloved by

1865
01:09:07,359 --> 01:09:11,839
list programmers especially macros very

1866
01:09:09,759 --> 01:09:14,159
much loved by list programmers um

1867
01:09:11,839 --> 01:09:16,400
and if they're so great why aren't they

1868
01:09:14,159 --> 01:09:18,560
in other languages as much uh and i

1869
01:09:16,400 --> 01:09:20,799
think in my opinion at least it's mostly

1870
01:09:18,560 --> 01:09:22,560
because lisp has such a beautiful like

1871
01:09:20,799 --> 01:09:24,640
regular syntax

1872
01:09:22,560 --> 01:09:26,719
uh that it makes it a really easy

1873
01:09:24,640 --> 01:09:28,239
interface to be manipulating through the

1874
01:09:26,719 --> 01:09:30,480
meta programming right

1875
01:09:28,239 --> 01:09:32,400
you know it's all lists embed lists and

1876
01:09:30,480 --> 01:09:34,480
lists you know just recursive lists it's

1877
01:09:32,400 --> 01:09:36,719
very easy to iterate through that kind

1878
01:09:34,480 --> 01:09:38,640
of stuff same thing with fourth um

1879
01:09:36,719 --> 01:09:40,319
and that's why

1880
01:09:38,640 --> 01:09:42,000
it works so well and i think that's

1881
01:09:40,319 --> 01:09:43,040
where a lot of the libraries that try to

1882
01:09:42,000 --> 01:09:45,040
implement this kind of stuff in

1883
01:09:43,040 --> 01:09:46,400
languages that weren't designed for it

1884
01:09:45,040 --> 01:09:48,319
like fall down

1885
01:09:46,400 --> 01:09:50,000
because they're trying to use the ast

1886
01:09:48,319 --> 01:09:52,640
representation that lisp uses in other

1887
01:09:50,000 --> 01:09:53,920
languages with a really complex ast

1888
01:09:52,640 --> 01:09:55,520
and it makes it you know just makes it

1889
01:09:53,920 --> 01:09:57,520
confusing and hard to

1890
01:09:55,520 --> 01:09:58,719
reason about in my opinion at least um

1891
01:09:57,520 --> 01:10:01,600
and and so that's what i was trying to

1892
01:09:58,719 --> 01:10:04,159
do with this project was to capture that

1893
01:10:01,600 --> 01:10:06,320
simplicity in the interface that lisp

1894
01:10:04,159 --> 01:10:07,920
has that makes it so good

1895
01:10:06,320 --> 01:10:11,520
and so yeah speaking of we're going to

1896
01:10:07,920 --> 01:10:11,520
get into bytecode macro

1897
01:10:11,600 --> 01:10:16,480
so

1898
01:10:12,320 --> 01:10:19,360
how bytecode macro works is it takes you

1899
01:10:16,480 --> 01:10:21,840
know it's a decorator that takes a

1900
01:10:19,360 --> 01:10:24,080
macro function and the function to apply

1901
01:10:21,840 --> 01:10:25,840
that macro to

1902
01:10:24,080 --> 01:10:28,320
converts that

1903
01:10:25,840 --> 01:10:32,159
function it passed in this case test to

1904
01:10:28,320 --> 01:10:34,159
a list of tuples or tuples

1905
01:10:32,159 --> 01:10:36,480
uh and

1906
01:10:34,159 --> 01:10:38,800
you know those tuples are instructions

1907
01:10:36,480 --> 01:10:40,560
in the bytecode in in the in the buy

1908
01:10:38,800 --> 01:10:41,920
code right it's a simplified human

1909
01:10:40,560 --> 01:10:44,320
readable version

1910
01:10:41,920 --> 01:10:45,600
of that by code and it passes this list

1911
01:10:44,320 --> 01:10:46,880
of tuples

1912
01:10:45,600 --> 01:10:49,679
to the

1913
01:10:46,880 --> 01:10:53,120
macro function in this case macro trace

1914
01:10:49,679 --> 01:10:56,400
uh and then that macro function can

1915
01:10:53,120 --> 01:10:58,719
modify that by code that linear list of

1916
01:10:56,400 --> 01:11:00,640
tuples to be whatever it wanted um

1917
01:10:58,719 --> 01:11:03,199
so in this case we've changed hello

1918
01:11:00,640 --> 01:11:05,679
world to hello friends

1919
01:11:03,199 --> 01:11:07,600
and then you've returned this newly

1920
01:11:05,679 --> 01:11:09,840
modified list of tuples

1921
01:11:07,600 --> 01:11:11,280
and it will execute this as if it was

1922
01:11:09,840 --> 01:11:13,679
the original function

1923
01:11:11,280 --> 01:11:15,280
um and there we go this is saying hello

1924
01:11:13,679 --> 01:11:17,120
friends at the end

1925
01:11:15,280 --> 01:11:18,480
um and this

1926
01:11:17,120 --> 01:11:20,159
i think that's really cool and that's

1927
01:11:18,480 --> 01:11:21,440
that's how it works is very simple i

1928
01:11:20,159 --> 01:11:23,520
think it's you know easy enough to

1929
01:11:21,440 --> 01:11:25,040
understand and this is the

1930
01:11:23,520 --> 01:11:27,520
macro that was used to produce the

1931
01:11:25,040 --> 01:11:30,560
visualizations we saw just

1932
01:11:27,520 --> 01:11:32,719
last slide uh it takes in the tuples and

1933
01:11:30,560 --> 01:11:35,840
then 4n in the tuples of prince enso

1934
01:11:32,719 --> 01:11:37,360
prints each tuple in a line

1935
01:11:35,840 --> 01:11:38,080
and then it returns it without changing

1936
01:11:37,360 --> 01:11:40,480
it

1937
01:11:38,080 --> 01:11:43,040
so that's how we can see you know

1938
01:11:40,480 --> 01:11:44,800
what the type of representation of the

1939
01:11:43,040 --> 01:11:46,480
above function is

1940
01:11:44,800 --> 01:11:48,080
uh

1941
01:11:46,480 --> 01:11:49,920
and you know we can move on here's the

1942
01:11:48,080 --> 01:11:51,679
go to example we saw right at the side

1943
01:11:49,920 --> 01:11:53,840
that i thought was really cool

1944
01:11:51,679 --> 01:11:55,280
um just take a little bit it doesn't

1945
01:11:53,840 --> 01:11:56,800
matter too much it just works like a

1946
01:11:55,280 --> 01:11:58,800
go-to right

1947
01:11:56,800 --> 01:12:00,080
uh and here's the macro implementation

1948
01:11:58,800 --> 01:12:01,760
looks a little scary but it's really

1949
01:12:00,080 --> 01:12:04,400
simple um

1950
01:12:01,760 --> 01:12:06,000
and so what it does is it checks if

1951
01:12:04,400 --> 01:12:08,480
there is a string

1952
01:12:06,000 --> 01:12:11,040
which has goto in it

1953
01:12:08,480 --> 01:12:13,679
uh and then if it does it replaces the

1954
01:12:11,040 --> 01:12:15,520
string loading command with

1955
01:12:13,679 --> 01:12:17,920
a command that

1956
01:12:15,520 --> 01:12:19,520
a jump command here with wherever the go

1957
01:12:17,920 --> 01:12:21,360
to was targeting

1958
01:12:19,520 --> 01:12:23,040
and if there is a string

1959
01:12:21,360 --> 01:12:25,360
which has a

1960
01:12:23,040 --> 01:12:27,360
colon at the end or like a label then it

1961
01:12:25,360 --> 01:12:28,640
replaces that instruction with a label

1962
01:12:27,360 --> 01:12:31,920
instruction

1963
01:12:28,640 --> 01:12:34,239
um with whatever the label was

1964
01:12:31,920 --> 01:12:36,560
and then otherwise it's just you know

1965
01:12:34,239 --> 01:12:39,280
appends through normally and that's how

1966
01:12:36,560 --> 01:12:41,920
the go-to macro works i think it works

1967
01:12:39,280 --> 01:12:43,600
so simply because there i added um

1968
01:12:41,920 --> 01:12:46,320
it because there is

1969
01:12:43,600 --> 01:12:48,320
jump commands in

1970
01:12:46,320 --> 01:12:50,159
uh the original byte code and it is

1971
01:12:48,320 --> 01:12:52,239
worth mentioning that this isn't an

1972
01:12:50,159 --> 01:12:53,920
exact one to one of the original by code

1973
01:12:52,239 --> 01:12:55,280
in human readable form i've made a few

1974
01:12:53,920 --> 01:12:58,080
small changes

1975
01:12:55,280 --> 01:12:59,760
just to make it better um but yeah it's

1976
01:12:58,080 --> 01:13:01,360
pretty cool i think

1977
01:12:59,760 --> 01:13:02,719
uh and then here's the last one which is

1978
01:13:01,360 --> 01:13:04,719
a little complicated it took me a while

1979
01:13:02,719 --> 01:13:07,360
to get my head around after i wrote it

1980
01:13:04,719 --> 01:13:09,280
uh what it does it is just a

1981
01:13:07,360 --> 01:13:11,120
an example to try to show that it's got

1982
01:13:09,280 --> 01:13:12,719
so much power and you can really do

1983
01:13:11,120 --> 01:13:14,000
anything you want with this you don't

1984
01:13:12,719 --> 01:13:16,320
have to do something that already exists

1985
01:13:14,000 --> 01:13:19,040
you can let your mind run through

1986
01:13:16,320 --> 01:13:21,760
so to speak um and so heads one works is

1987
01:13:19,040 --> 01:13:23,679
within the join and block

1988
01:13:21,760 --> 01:13:26,560
it's it takes the output of the previous

1989
01:13:23,679 --> 01:13:27,760
line and feeds it in as the first input

1990
01:13:26,560 --> 01:13:30,719
on the

1991
01:13:27,760 --> 01:13:32,080
of the function on the next line um and

1992
01:13:30,719 --> 01:13:33,600
and n equals three doesn't return

1993
01:13:32,080 --> 01:13:35,600
anything in python

1994
01:13:33,600 --> 01:13:38,080
uh but i've made it return like a walrus

1995
01:13:35,600 --> 01:13:38,880
operator just to make it uh just to make

1996
01:13:38,080 --> 01:13:41,120
this

1997
01:13:38,880 --> 01:13:42,880
this whole thing work but so for this

1998
01:13:41,120 --> 01:13:44,960
example right it goes from range three

1999
01:13:42,880 --> 01:13:48,080
to six and then it sums that range from

2000
01:13:44,960 --> 01:13:49,840
three to six and then it times it by two

2001
01:13:48,080 --> 01:13:52,400
and then it prints that output and which

2002
01:13:49,840 --> 01:13:53,520
does print 24 and you can see the maths

2003
01:13:52,400 --> 01:13:54,480
just here

2004
01:13:53,520 --> 01:13:55,840
um

2005
01:13:54,480 --> 01:13:57,840
and i think that's really cool this is

2006
01:13:55,840 --> 01:13:59,120
just you know i think that's it's pretty

2007
01:13:57,840 --> 01:14:02,080
crazy

2008
01:13:59,120 --> 01:14:04,719
so uh just miscellaneous

2009
01:14:02,080 --> 01:14:06,880
should you use this in your code um i i

2010
01:14:04,719 --> 01:14:08,320
would i hope you wouldn't i think it's

2011
01:14:06,880 --> 01:14:10,080
you know it's really bad meta

2012
01:14:08,320 --> 01:14:11,920
programming a lot of the time is

2013
01:14:10,080 --> 01:14:13,360
not optimal to using your code at all it

2014
01:14:11,920 --> 01:14:14,800
can lead to some very hard to read code

2015
01:14:13,360 --> 01:14:17,440
which we see before

2016
01:14:14,800 --> 01:14:21,040
um and i think fundamentally

2017
01:14:17,440 --> 01:14:22,719
it's it's kind of anti-pythonic it's not

2018
01:14:21,040 --> 01:14:24,560
it's it's against the principles of

2019
01:14:22,719 --> 01:14:26,719
python um and

2020
01:14:24,560 --> 01:14:29,760
so i've got the zener python here

2021
01:14:26,719 --> 01:14:32,800
the of python the python poem that

2022
01:14:29,760 --> 01:14:35,360
explains exactly

2023
01:14:32,800 --> 01:14:38,400
uh the the core design principles of

2024
01:14:35,360 --> 01:14:41,360
python and i think you can see that it's

2025
01:14:38,400 --> 01:14:43,040
it's just against most of it right um

2026
01:14:41,360 --> 01:14:45,199
you know explicit is better than

2027
01:14:43,040 --> 01:14:48,080
implicit that's that's the whole idea

2028
01:14:45,199 --> 01:14:51,440
isn't it right to try to build

2029
01:14:48,080 --> 01:14:53,040
uh implicit changes to python

2030
01:14:51,440 --> 01:14:54,800
and then you know

2031
01:14:53,040 --> 01:14:56,560
what's the other one oh yeah

2032
01:14:54,800 --> 01:14:58,400
if the implementation is hard to explain

2033
01:14:56,560 --> 01:14:59,920
it's a bad idea i think that's that's a

2034
01:14:58,400 --> 01:15:01,199
given right i spent a whole 10 minutes

2035
01:14:59,920 --> 01:15:03,840
trying to explain it

2036
01:15:01,199 --> 01:15:05,600
um and then the last one being special

2037
01:15:03,840 --> 01:15:06,960
cases aren't special enough to break the

2038
01:15:05,600 --> 01:15:09,040
rules right that's the whole point we're

2039
01:15:06,960 --> 01:15:11,199
trying to break the rules of python in

2040
01:15:09,040 --> 01:15:13,120
order to account for these special cases

2041
01:15:11,199 --> 01:15:15,600
right um

2042
01:15:13,120 --> 01:15:18,080
and yeah so then you're probably asking

2043
01:15:15,600 --> 01:15:20,000
why did you do this then you know what's

2044
01:15:18,080 --> 01:15:21,520
what's the whole point and i think you

2045
01:15:20,000 --> 01:15:23,199
know you can do the stuff that list

2046
01:15:21,520 --> 01:15:25,120
macros allow you to do you can extend

2047
01:15:23,199 --> 01:15:26,560
python syntax you can even write domain

2048
01:15:25,120 --> 01:15:28,880
specific languages if you really wanted

2049
01:15:26,560 --> 01:15:30,080
to there's some element of code reuses

2050
01:15:28,880 --> 01:15:31,920
but i think

2051
01:15:30,080 --> 01:15:33,920
at the end of the day it's just a lot of

2052
01:15:31,920 --> 01:15:35,360
fun it was fun to make and i think it's

2053
01:15:33,920 --> 01:15:36,560
fun to mess around with the internals of

2054
01:15:35,360 --> 01:15:38,000
python

2055
01:15:36,560 --> 01:15:40,080
and it's fun to extend language like

2056
01:15:38,000 --> 01:15:41,760
this i think it's you know i i enjoyed i

2057
01:15:40,080 --> 01:15:43,199
enjoyed doing it and you know i hope you

2058
01:15:41,760 --> 01:15:45,760
would enjoy messing around with the

2059
01:15:43,199 --> 01:15:47,920
library a little bit uh and yeah so

2060
01:15:45,760 --> 01:15:49,840
that's the end of my talk here are the

2061
01:15:47,920 --> 01:15:51,120
links around the biker so if you want to

2062
01:15:49,840 --> 01:15:53,440
check out the github

2063
01:15:51,120 --> 01:15:54,800
you can here's a link

2064
01:15:53,440 --> 01:15:57,679
and if you want to just install the

2065
01:15:54,800 --> 01:15:59,120
library go ahead uh you know just do it

2066
01:15:57,679 --> 01:16:00,080
instantly like this

2067
01:15:59,120 --> 01:16:03,440
there's the

2068
01:16:00,080 --> 01:16:04,800
installation and command it's on pip

2069
01:16:03,440 --> 01:16:06,640
but i would recommend reading through

2070
01:16:04,800 --> 01:16:08,320
the documentation on github if you do

2071
01:16:06,640 --> 01:16:09,360
want to give this library a little bit

2072
01:16:08,320 --> 01:16:10,960
of a try

2073
01:16:09,360 --> 01:16:13,520
uh thank you

2074
01:16:10,960 --> 01:16:16,080
fantastic thanks so much

2075
01:16:13,520 --> 01:16:18,400
our next talk is a group talk we'll be

2076
01:16:16,080 --> 01:16:20,719
hearing from amelia dalia vyan and

2077
01:16:18,400 --> 01:16:22,560
elizabeth about building cooker berry

2078
01:16:20,719 --> 01:16:24,239
apps for learning and they've been

2079
01:16:22,560 --> 01:16:25,280
programming just since the beginning of

2080
01:16:24,239 --> 01:16:27,440
this year

2081
01:16:25,280 --> 01:16:30,159
when i asked their favorite emoji it was

2082
01:16:27,440 --> 01:16:33,600
a 50 50 split between the rolling eyes

2083
01:16:30,159 --> 01:16:35,679
emoji and the sun emoji a dangerous mix

2084
01:16:33,600 --> 01:16:37,120
i hope you've got sunnies too

2085
01:16:35,679 --> 01:16:38,719
when i asked what they recently learned

2086
01:16:37,120 --> 01:16:40,960
about computing they said they've been

2087
01:16:38,719 --> 01:16:43,520
learning about caesar ciphers encoding

2088
01:16:40,960 --> 01:16:46,159
and decoding and also about the general

2089
01:16:43,520 --> 01:16:47,440
concept of computers not having a human

2090
01:16:46,159 --> 01:16:49,760
language

2091
01:16:47,440 --> 01:16:51,360
it's a deep rabbit hole to go down but

2092
01:16:49,760 --> 01:16:54,159
an interesting one

2093
01:16:51,360 --> 01:16:54,159
let's hear their talk

2094
01:16:54,239 --> 01:16:57,520
hi i'm emilia

2095
01:16:56,159 --> 01:16:58,960
elizabeth

2096
01:16:57,520 --> 01:17:00,560
leon

2097
01:16:58,960 --> 01:17:03,280
dalia

2098
01:17:00,560 --> 01:17:04,719
and we're in year 9 at plc sydney

2099
01:17:03,280 --> 01:17:06,400
and we are the cambridge computer

2100
01:17:04,719 --> 01:17:08,880
science class

2101
01:17:06,400 --> 01:17:10,640
i've been coding for about seven months

2102
01:17:08,880 --> 01:17:12,719
started this year

2103
01:17:10,640 --> 01:17:14,719
since the start of this year

2104
01:17:12,719 --> 01:17:16,560
around seven months

2105
01:17:14,719 --> 01:17:18,480
i got involved in computer science

2106
01:17:16,560 --> 01:17:20,239
because i believe that if i learn about

2107
01:17:18,480 --> 01:17:22,320
something that keeps evolving it will

2108
01:17:20,239 --> 01:17:23,760
help me in the future

2109
01:17:22,320 --> 01:17:24,960
i think it is important to know about

2110
01:17:23,760 --> 01:17:27,760
because the time we are living in

2111
01:17:24,960 --> 01:17:29,280
revolves more and more around computers

2112
01:17:27,760 --> 01:17:31,600
i believe that computer science will be

2113
01:17:29,280 --> 01:17:33,199
a helpful tool in our future

2114
01:17:31,600 --> 01:17:35,120
and i thought it would be interesting to

2115
01:17:33,199 --> 01:17:37,199
learn

2116
01:17:35,120 --> 01:17:38,719
for our assignment our teacher mrs noble

2117
01:17:37,199 --> 01:17:40,320
challenged us to design and code

2118
01:17:38,719 --> 01:17:41,520
programs in python that will help with

2119
01:17:40,320 --> 01:17:43,360
learning

2120
01:17:41,520 --> 01:17:45,679
to complete the challenge we got to use

2121
01:17:43,360 --> 01:17:47,360
cooker berries an embedded device coded

2122
01:17:45,679 --> 01:17:49,440
in micro python

2123
01:17:47,360 --> 01:17:52,159
they have lots of inputs and outputs

2124
01:17:49,440 --> 01:17:55,360
like buttons screens and speakers that

2125
01:17:52,159 --> 01:17:57,040
we could use to create our programs

2126
01:17:55,360 --> 01:17:59,040
together we came up with seven apps to

2127
01:17:57,040 --> 01:18:02,800
help with practicing music and studying

2128
01:17:59,040 --> 01:18:02,800
for school that we'll show you today

2129
01:18:05,600 --> 01:18:09,360
i created two apps to help with language

2130
01:18:07,280 --> 01:18:10,719
learning and organization i chose to

2131
01:18:09,360 --> 01:18:12,080
create programs for this area of

2132
01:18:10,719 --> 01:18:13,520
learning because i personally struggle

2133
01:18:12,080 --> 01:18:15,600
with organizing myself and finding

2134
01:18:13,520 --> 01:18:16,960
myself wanting to study i started by

2135
01:18:15,600 --> 01:18:18,000
thinking about what i would want in a

2136
01:18:16,960 --> 01:18:19,360
learning app and thought it would be

2137
01:18:18,000 --> 01:18:21,360
good if i could use technology to

2138
01:18:19,360 --> 01:18:23,760
emulate that so i decided to make my

2139
01:18:21,360 --> 01:18:25,440
calendar and flashcard cookieberry apps

2140
01:18:23,760 --> 01:18:27,120
my first app was my calendar app and i

2141
01:18:25,440 --> 01:18:29,199
started off by working on the design and

2142
01:18:27,120 --> 01:18:30,719
features i wanted my original idea was

2143
01:18:29,199 --> 01:18:32,320
to have a large timetable template

2144
01:18:30,719 --> 01:18:34,159
similar to a school timetable with all

2145
01:18:32,320 --> 01:18:35,440
the days of the week however i realized

2146
01:18:34,159 --> 01:18:36,960
this would not be possible since the

2147
01:18:35,440 --> 01:18:38,880
kookaburra pixels are quite large and

2148
01:18:36,960 --> 01:18:40,239
there would not be enough space i came

2149
01:18:38,880 --> 01:18:42,239
up with a new design that fit the screen

2150
01:18:40,239 --> 01:18:44,560
which only showed one day at a time once

2151
01:18:42,239 --> 01:18:46,080
i had my design i needed to start coding

2152
01:18:44,560 --> 01:18:48,159
i first coded the lines that would make

2153
01:18:46,080 --> 01:18:49,440
up the timetable boxes there's a lot of

2154
01:18:48,159 --> 01:18:50,800
python to learn because this is the

2155
01:18:49,440 --> 01:18:52,159
first time i've encountered code for

2156
01:18:50,800 --> 01:18:54,159
displaying boxes and shapes on the

2157
01:18:52,159 --> 01:18:55,679
kookaberry getting it to look like my

2158
01:18:54,159 --> 01:18:56,960
design took a lot of trial and error i

2159
01:18:55,679 --> 01:18:58,239
also had a lot of lists of data to

2160
01:18:56,960 --> 01:19:00,800
organize to ensure that different study

2161
01:18:58,239 --> 01:19:02,320
sessions would associate with their day

2162
01:19:00,800 --> 01:19:03,920
i also needed to do things like at the

2163
01:19:02,320 --> 01:19:05,199
week to loop back round to the start

2164
01:19:03,920 --> 01:19:07,520
when using buttons to click through the

2165
01:19:05,199 --> 01:19:09,280
days in the timetable i needed to use a

2166
01:19:07,520 --> 01:19:10,800
special case when i reached the end of

2167
01:19:09,280 --> 01:19:12,159
the week so i could get back to the data

2168
01:19:10,800 --> 01:19:14,239
stored in my list for the start of the

2169
01:19:12,159 --> 01:19:16,400
week when i run my app it opens to

2170
01:19:14,239 --> 01:19:18,320
sunday and the task that day to go the

2171
01:19:16,400 --> 01:19:19,840
next day click d to go the previous day

2172
01:19:18,320 --> 01:19:21,280
click c

2173
01:19:19,840 --> 01:19:22,719
for my second app i thought about how

2174
01:19:21,280 --> 01:19:24,400
flashcards are helpful tool when

2175
01:19:22,719 --> 01:19:25,360
learning so i thought a flashcard app

2176
01:19:24,400 --> 01:19:27,840
would be a good choice for this

2177
01:19:25,360 --> 01:19:29,199
assignment i wanted to be able to show a

2178
01:19:27,840 --> 01:19:30,800
word in one language and then have the

2179
01:19:29,199 --> 01:19:33,280
user click to find out and memorize the

2180
01:19:30,800 --> 01:19:34,880
translation it ended up working so if a

2181
01:19:33,280 --> 01:19:36,880
is pressed it will display a word to

2182
01:19:34,880 --> 01:19:39,120
memorize in this case i chose various

2183
01:19:36,880 --> 01:19:41,920
french words if b is pressed the english

2184
01:19:39,120 --> 01:19:43,600
is shown if d is pressed the next card

2185
01:19:41,920 --> 01:19:45,280
is shown with the english option i

2186
01:19:43,600 --> 01:19:46,880
wanted to make sure that it would be

2187
01:19:45,280 --> 01:19:48,320
able to go backwards through the cards

2188
01:19:46,880 --> 01:19:50,239
as well but i couldn't figure out how

2189
01:19:48,320 --> 01:19:51,760
with the time i had when the final words

2190
01:19:50,239 --> 01:19:53,520
are reached and d is pressed to go to

2191
01:19:51,760 --> 01:19:55,120
the next one a congratulatory message

2192
01:19:53,520 --> 01:19:57,199
appears as well as the vocab that should

2193
01:19:55,120 --> 01:20:00,159
be memorized the user has the option to

2194
01:19:57,199 --> 01:20:01,440
repeat the system by clicking a

2195
01:20:00,159 --> 01:20:02,560
the hardest part for this app was

2196
01:20:01,440 --> 01:20:03,920
finding a way to switch between the

2197
01:20:02,560 --> 01:20:06,159
cards because there were multiple bits

2198
01:20:03,920 --> 01:20:08,400
of data associated with each screen

2199
01:20:06,159 --> 01:20:10,000
i really enjoyed learning about these

2200
01:20:08,400 --> 01:20:11,679
new ways to use python because i never

2201
01:20:10,000 --> 01:20:13,120
encountered them before

2202
01:20:11,679 --> 01:20:14,960
in the future i hope to utilize what

2203
01:20:13,120 --> 01:20:18,840
i've learned so i can find ways to

2204
01:20:14,960 --> 01:20:18,840
improve my programming skills

2205
01:20:20,080 --> 01:20:24,560
i created an app to help with studying

2206
01:20:22,800 --> 01:20:26,560
i chose to create a program for this

2207
01:20:24,560 --> 01:20:27,920
area of learning because it helps lots

2208
01:20:26,560 --> 01:20:30,800
of people when they are studying for

2209
01:20:27,920 --> 01:20:32,719
multiple subjects at one given time

2210
01:20:30,800 --> 01:20:34,719
i started out by thinking about how i

2211
01:20:32,719 --> 01:20:36,480
could use programming and the features

2212
01:20:34,719 --> 01:20:39,360
of the kookaburry like the buttons and

2213
01:20:36,480 --> 01:20:42,560
screen to help students learn

2214
01:20:39,360 --> 01:20:44,800
and so i decided to make a pop quiz app

2215
01:20:42,560 --> 01:20:46,719
to do this i first had to come up with a

2216
01:20:44,800 --> 01:20:48,960
plan of what questions to put in the

2217
01:20:46,719 --> 01:20:52,080
quiz as well as coming up with both the

2218
01:20:48,960 --> 01:20:54,800
correct and incorrect answers

2219
01:20:52,080 --> 01:20:57,199
to code the actual quiz i had to use a

2220
01:20:54,800 --> 01:20:59,520
lot of lists for storing the question

2221
01:20:57,199 --> 01:21:01,040
the multiple choice answers as well as a

2222
01:20:59,520 --> 01:21:03,280
list that says which is the correct

2223
01:21:01,040 --> 01:21:05,120
answer for each question

2224
01:21:03,280 --> 01:21:06,719
getting it to work was tricky because

2225
01:21:05,120 --> 01:21:09,280
sometimes the question wouldn't match up

2226
01:21:06,719 --> 01:21:11,199
with the corresponding answer but once i

2227
01:21:09,280 --> 01:21:13,840
made sure my lists were in the right

2228
01:21:11,199 --> 01:21:16,000
order of answers and questions and that

2229
01:21:13,840 --> 01:21:17,280
i was using the right indexes it all

2230
01:21:16,000 --> 01:21:19,280
worked

2231
01:21:17,280 --> 01:21:22,000
along the way i changed the design of my

2232
01:21:19,280 --> 01:21:23,600
project like including more questions

2233
01:21:22,000 --> 01:21:25,360
since i thought this would be useful to

2234
01:21:23,600 --> 01:21:28,800
give the user more questions to test

2235
01:21:25,360 --> 01:21:30,880
themselves on

2236
01:21:28,800 --> 01:21:33,280
in the end i got an app that works by

2237
01:21:30,880 --> 01:21:35,520
pressing a to start it and then when the

2238
01:21:33,280 --> 01:21:39,199
question and answer options appear you

2239
01:21:35,520 --> 01:21:41,040
select an answer using buttons a to d

2240
01:21:39,199 --> 01:21:42,800
once you have chosen your answer it

2241
01:21:41,040 --> 01:21:45,360
tells you if you are correct or not and

2242
01:21:42,800 --> 01:21:47,280
lets you proceed to the next question

2243
01:21:45,360 --> 01:21:49,280
in the future i hope to code this app

2244
01:21:47,280 --> 01:21:51,440
again so i can improve some features of

2245
01:21:49,280 --> 01:21:53,760
it like more questions and maybe a score

2246
01:21:51,440 --> 01:21:57,400
tally at the end and continue adding to

2247
01:21:53,760 --> 01:21:57,400
it to make it better

2248
01:22:01,280 --> 01:22:05,199
to do this project i started out by

2249
01:22:03,280 --> 01:22:06,960
thinking about what activities i do for

2250
01:22:05,199 --> 01:22:09,280
school and who could help with my own

2251
01:22:06,960 --> 01:22:11,520
learning i play a musical instrument in

2252
01:22:09,280 --> 01:22:13,280
school so i decided to make an app that

2253
01:22:11,520 --> 01:22:14,480
would make it easier to practice music

2254
01:22:13,280 --> 01:22:16,639
from home

2255
01:22:14,480 --> 01:22:18,880
so i decided to make two apps a

2256
01:22:16,639 --> 01:22:20,560
metronome that helps you play in time

2257
01:22:18,880 --> 01:22:22,880
and an app that keeps track of how many

2258
01:22:20,560 --> 01:22:24,719
times you've practiced different songs

2259
01:22:22,880 --> 01:22:26,400
for my metronome app i used a cooker

2260
01:22:24,719 --> 01:22:27,280
berry with the speaker module to play

2261
01:22:26,400 --> 01:22:29,520
the beat

2262
01:22:27,280 --> 01:22:31,280
to code my app i had to think about how

2263
01:22:29,520 --> 01:22:32,800
people would interact with it and i

2264
01:22:31,280 --> 01:22:35,120
decided to use four buttons on the

2265
01:22:32,800 --> 01:22:36,840
cooker berry for different functions on

2266
01:22:35,120 --> 01:22:39,520
off faster and

2267
01:22:36,840 --> 01:22:42,080
slower to reuse the metronome you click

2268
01:22:39,520 --> 01:22:44,400
a to start it once a is clicked the beat

2269
01:22:42,080 --> 01:22:46,639
will start coming from the speaker if

2270
01:22:44,400 --> 01:22:48,880
you click c the tempo will become slower

2271
01:22:46,639 --> 01:22:53,840
and if you click d the tempo will become

2272
01:22:48,880 --> 01:22:53,840
faster to stop the metronome click b

2273
01:22:53,920 --> 01:22:58,880
because i wanted to be able to vary the

2274
01:22:56,080 --> 01:23:00,880
speed of my metronome gaining the timing

2275
01:22:58,880 --> 01:23:03,120
was tricky i had to have two different

2276
01:23:00,880 --> 01:23:04,880
time values so that it would stay in b

2277
01:23:03,120 --> 01:23:07,040
for the length of the b and the time

2278
01:23:04,880 --> 01:23:09,280
between beats but once i understood how

2279
01:23:07,040 --> 01:23:11,280
to get the time right i could add two or

2280
01:23:09,280 --> 01:23:14,159
subtract from the variables to change

2281
01:23:11,280 --> 01:23:16,320
the tempo to get it working

2282
01:23:14,159 --> 01:23:17,679
once i had my metronome working i wanted

2283
01:23:16,320 --> 01:23:19,920
a way to track the songs i was

2284
01:23:17,679 --> 01:23:21,679
practicing i started out by coding a

2285
01:23:19,920 --> 01:23:24,400
list of all the names of the songs i

2286
01:23:21,679 --> 01:23:26,560
wanted to track to practice on

2287
01:23:24,400 --> 01:23:28,320
once i had done that i created another

2288
01:23:26,560 --> 01:23:30,480
list that would keep a tally for each

2289
01:23:28,320 --> 01:23:33,199
song so that we would have a different

2290
01:23:30,480 --> 01:23:34,639
count for each songs in the first list

2291
01:23:33,199 --> 01:23:36,560
getting the tally to match up with

2292
01:23:34,639 --> 01:23:38,639
correct song was tricky because it would

2293
01:23:36,560 --> 01:23:40,880
keep the same tally but then go to the

2294
01:23:38,639 --> 01:23:43,760
next song but once i worked out the

2295
01:23:40,880 --> 01:23:46,080
indexes i got it working

2296
01:23:43,760 --> 01:23:48,800
my finished app looked like this

2297
01:23:46,080 --> 01:23:50,880
you can click a and b to move backwards

2298
01:23:48,800 --> 01:23:52,159
and forwards between different songs and

2299
01:23:50,880 --> 01:23:53,520
click c and d

2300
01:23:52,159 --> 01:23:56,159
to add or

2301
01:23:53,520 --> 01:23:57,920
take away from the tally i even added an

2302
01:23:56,159 --> 01:24:00,480
extra feature where once the tally

2303
01:23:57,920 --> 01:24:02,400
reaches five on the song this neopixel

2304
01:24:00,480 --> 01:24:05,600
light display will flash to congratulate

2305
01:24:02,400 --> 01:24:07,760
you for reaching your goal

2306
01:24:05,600 --> 01:24:09,760
i really enjoyed making my apps because

2307
01:24:07,760 --> 01:24:11,760
i could put my skills to use and create

2308
01:24:09,760 --> 01:24:15,080
something of my own rather than copying

2309
01:24:11,760 --> 01:24:15,080
from a booklet

2310
01:24:18,000 --> 01:24:21,840
i created two wraps to help with study i

2311
01:24:20,239 --> 01:24:23,520
chose to create programs for this area

2312
01:24:21,840 --> 01:24:25,440
of learning because i do not have a good

2313
01:24:23,520 --> 01:24:27,600
study routine and never feel motivated

2314
01:24:25,440 --> 01:24:29,360
to study

2315
01:24:27,600 --> 01:24:30,800
i started out by thinking about the ways

2316
01:24:29,360 --> 01:24:32,639
that have helped me to study and thought

2317
01:24:30,800 --> 01:24:34,960
it'd be good if i could use technology

2318
01:24:32,639 --> 01:24:37,280
to make study time and quota

2319
01:24:34,960 --> 01:24:39,600
so first i decided to make an automatic

2320
01:24:37,280 --> 01:24:42,000
pomodoro method timer which is a method

2321
01:24:39,600 --> 01:24:44,560
that is 25 minutes of study and a five

2322
01:24:42,000 --> 01:24:46,480
minute break and this continues

2323
01:24:44,560 --> 01:24:48,719
and once i had done that i made an app

2324
01:24:46,480 --> 01:24:50,480
that displays inspirational quotes to

2325
01:24:48,719 --> 01:24:52,159
keep me motivated

2326
01:24:50,480 --> 01:24:54,400
the first thing i did was work out how

2327
01:24:52,159 --> 01:24:56,400
my study time should work my first

2328
01:24:54,400 --> 01:24:58,159
version of the project was an app that

2329
01:24:56,400 --> 01:24:59,840
would start when you press button a and

2330
01:24:58,159 --> 01:25:01,920
it would start counting down from 25

2331
01:24:59,840 --> 01:25:03,840
minutes for the study session and once

2332
01:25:01,920 --> 01:25:06,159
that finished it would count down from 5

2333
01:25:03,840 --> 01:25:08,239
minutes for the break time

2334
01:25:06,159 --> 01:25:10,560
i then added on extra features like

2335
01:25:08,239 --> 01:25:12,159
doing multiple sessions in a row adding

2336
01:25:10,560 --> 01:25:14,960
a speaker to make sounds at the end of

2337
01:25:12,159 --> 01:25:16,719
each set each session the ability to end

2338
01:25:14,960 --> 01:25:18,880
the session and account for how many

2339
01:25:16,719 --> 01:25:20,639
sessions you've done in a row

2340
01:25:18,880 --> 01:25:22,560
the trickiest thing about making the

2341
01:25:20,639 --> 01:25:24,480
time was that the code got really long

2342
01:25:22,560 --> 01:25:26,320
by the end and i needed to keep adding

2343
01:25:24,480 --> 01:25:28,320
to and editing lots of different parts

2344
01:25:26,320 --> 01:25:30,400
of the code to create my new features

2345
01:25:28,320 --> 01:25:32,239
because it was all connected but with

2346
01:25:30,400 --> 01:25:33,520
the help of my teacher i managed to work

2347
01:25:32,239 --> 01:25:34,480
out how to code it all and get it

2348
01:25:33,520 --> 01:25:36,560
working

2349
01:25:34,480 --> 01:25:38,320
my motivational quote was much more

2350
01:25:36,560 --> 01:25:40,560
straightforward to make it wasn't

2351
01:25:38,320 --> 01:25:42,560
particularly hard to code since i just

2352
01:25:40,560 --> 01:25:47,360
started to learn how to code it was

2353
01:25:42,560 --> 01:25:47,360
tricky to locate and fix syntax areas

2354
01:25:47,440 --> 01:25:51,520
using my app is also straightforward you

2355
01:25:49,440 --> 01:25:53,360
can just press button a this will make a

2356
01:25:51,520 --> 01:25:54,639
random code display on the screen of the

2357
01:25:53,360 --> 01:25:56,159
cooker berry

2358
01:25:54,639 --> 01:25:59,199
you can press button a every time you

2359
01:25:56,159 --> 01:26:01,199
want a new quote to read

2360
01:25:59,199 --> 01:26:02,800
i really enjoyed creating my own apps

2361
01:26:01,199 --> 01:26:05,760
because i was able to code without a

2362
01:26:02,800 --> 01:26:07,760
step-by-step guide in future i hope to

2363
01:26:05,760 --> 01:26:09,360
code more apps again so i can implement

2364
01:26:07,760 --> 01:26:11,440
what i've learned from this from the

2365
01:26:09,360 --> 01:26:14,639
beginning of this year up until now and

2366
01:26:11,440 --> 01:26:14,639
create even better apps

2367
01:26:17,360 --> 01:26:21,199
i really enjoyed making these apps

2368
01:26:19,199 --> 01:26:22,960
because it helps me understand what goes

2369
01:26:21,199 --> 01:26:24,880
on inside a computer when using a

2370
01:26:22,960 --> 01:26:26,800
similar app

2371
01:26:24,880 --> 01:26:28,639
using python for this project was good

2372
01:26:26,800 --> 01:26:30,320
because i learned most of the basics of

2373
01:26:28,639 --> 01:26:32,000
python at the beginning of the year so i

2374
01:26:30,320 --> 01:26:34,000
was able to understand the layout of the

2375
01:26:32,000 --> 01:26:36,320
code and how to code my apps for the

2376
01:26:34,000 --> 01:26:37,920
quicker berry

2377
01:26:36,320 --> 01:26:39,600
in the future i hope to use the cooker

2378
01:26:37,920 --> 01:26:41,840
barrier again so i can create more apps

2379
01:26:39,600 --> 01:26:43,840
with better coding

2380
01:26:41,840 --> 01:26:45,600
using python for these products was good

2381
01:26:43,840 --> 01:26:47,360
because we had spent the year learning

2382
01:26:45,600 --> 01:26:49,040
about python which is relatively easy

2383
01:26:47,360 --> 01:26:50,560
for a beginner and i look forward to

2384
01:26:49,040 --> 01:26:52,960
creating more projects with python in

2385
01:26:50,560 --> 01:26:52,960
the future

2386
01:26:56,719 --> 01:27:01,840
that was fantastic consoles look out

2387
01:26:59,679 --> 01:27:04,480
there are new games in town and here's

2388
01:27:01,840 --> 01:27:06,400
another one to talk about samuel is

2389
01:27:04,480 --> 01:27:09,840
talking about making an isometric game

2390
01:27:06,400 --> 01:27:11,679
from scratch in python using pygame

2391
01:27:09,840 --> 01:27:14,800
samuel's been programming for about six

2392
01:27:11,679 --> 01:27:16,560
years so since year six and his favorite

2393
01:27:14,800 --> 01:27:19,679
emoji is the

2394
01:27:16,560 --> 01:27:21,360
glasses emoji glasses represent

2395
01:27:19,679 --> 01:27:22,960
when i asked what he's recently learned

2396
01:27:21,360 --> 01:27:24,800
about computers he replied that

2397
01:27:22,960 --> 01:27:27,360
computers are bad at maths with

2398
01:27:24,800 --> 01:27:29,520
fractions floating point errors occur

2399
01:27:27,360 --> 01:27:32,080
when computers round numbers

2400
01:27:29,520 --> 01:27:34,000
and they are quite scary these errors

2401
01:27:32,080 --> 01:27:36,560
when they add up

2402
01:27:34,000 --> 01:27:38,159
let's hear more about his talk

2403
01:27:36,560 --> 01:27:40,400
good afternoon

2404
01:27:38,159 --> 01:27:43,199
welcome to my talk on how i implemented

2405
01:27:40,400 --> 01:27:44,880
isometric 3d in python's pie game

2406
01:27:43,199 --> 01:27:46,480
by way of introductions my name is

2407
01:27:44,880 --> 01:27:48,080
samuel price i go to san andreas

2408
01:27:46,480 --> 01:27:50,800
cathedral school and i've been

2409
01:27:48,080 --> 01:27:53,120
programming since grade six so about six

2410
01:27:50,800 --> 01:27:54,639
six years now and in those six years

2411
01:27:53,120 --> 01:27:57,840
something i've always really wanted to

2412
01:27:54,639 --> 01:27:59,440
do was develop a system for 3d

2413
01:27:57,840 --> 01:28:00,960
i've very much always been able to

2414
01:27:59,440 --> 01:28:02,880
believe that the best way to learn about

2415
01:28:00,960 --> 01:28:04,159
something is to do it yourself i started

2416
01:28:02,880 --> 01:28:06,639
off coding

2417
01:28:04,159 --> 01:28:08,320
some text-based rpgs moved on to 2d

2418
01:28:06,639 --> 01:28:09,920
games and i

2419
01:28:08,320 --> 01:28:11,199
made a clone of

2420
01:28:09,920 --> 01:28:14,480
quite a few

2421
01:28:11,199 --> 01:28:16,800
uh 2d games if you can think of one i've

2422
01:28:14,480 --> 01:28:19,600
probably done it

2423
01:28:16,800 --> 01:28:22,000
but one day i really really wanted to do

2424
01:28:19,600 --> 01:28:23,199
something in 3d the best system i could

2425
01:28:22,000 --> 01:28:26,480
think of

2426
01:28:23,199 --> 01:28:28,560
for me to implement was isometric 3d

2427
01:28:26,480 --> 01:28:30,960
now if you remember your math math

2428
01:28:28,560 --> 01:28:33,040
classes from grades one and two

2429
01:28:30,960 --> 01:28:34,880
what you get is these little series of

2430
01:28:33,040 --> 01:28:35,760
dots and you can connect them up to make

2431
01:28:34,880 --> 01:28:37,679
shapes

2432
01:28:35,760 --> 01:28:40,639
um essentially that works if you want to

2433
01:28:37,679 --> 01:28:42,719
travel say one unit in the x axis you

2434
01:28:40,639 --> 01:28:45,280
draw one of these lines in this

2435
01:28:42,719 --> 01:28:48,239
direction one of the y you do one of

2436
01:28:45,280 --> 01:28:49,840
these lines here one z one of those ones

2437
01:28:48,239 --> 01:28:51,520
and you chain those together to get

2438
01:28:49,840 --> 01:28:53,440
different points and to make two

2439
01:28:51,520 --> 01:28:55,440
different shapes

2440
01:28:53,440 --> 01:28:57,440
which all works pretty well and you can

2441
01:28:55,440 --> 01:28:58,719
see here that this is an implementation

2442
01:28:57,440 --> 01:29:01,040
of that model

2443
01:28:58,719 --> 01:29:02,159
it's just a little c where the ripples

2444
01:29:01,040 --> 01:29:04,000
come out

2445
01:29:02,159 --> 01:29:05,600
and you can you can see the depth you

2446
01:29:04,000 --> 01:29:07,280
can understand how it functions in three

2447
01:29:05,600 --> 01:29:08,400
dimensions you can see the rotation as

2448
01:29:07,280 --> 01:29:11,120
well

2449
01:29:08,400 --> 01:29:14,159
however the depth is also one problem of

2450
01:29:11,120 --> 01:29:17,040
this see in orthographic 3d which is

2451
01:29:14,159 --> 01:29:18,560
what we see in real life perspective 3d

2452
01:29:17,040 --> 01:29:20,320
when something's further away to your

2453
01:29:18,560 --> 01:29:22,800
smaller when it's closer you see it is

2454
01:29:20,320 --> 01:29:24,880
bigger but in isometric that's not quite

2455
01:29:22,800 --> 01:29:26,400
how it works if you have just a series

2456
01:29:24,880 --> 01:29:28,480
of dots it doesn't matter if you draw it

2457
01:29:26,400 --> 01:29:31,040
on one place or another it's always

2458
01:29:28,480 --> 01:29:33,600
going to be the same size

2459
01:29:31,040 --> 01:29:35,840
and so you can see here in this game a

2460
01:29:33,600 --> 01:29:38,960
small clone of temple run that i've

2461
01:29:35,840 --> 01:29:41,280
found a way to overcome this see

2462
01:29:38,960 --> 01:29:44,000
normally if you're jumping say from the

2463
01:29:41,280 --> 01:29:45,360
left most lane be hard to tell if you're

2464
01:29:44,000 --> 01:29:46,800
actually jumping from that lane or if

2465
01:29:45,360 --> 01:29:48,719
you're just in another lane staying

2466
01:29:46,800 --> 01:29:51,600
still

2467
01:29:48,719 --> 01:29:54,320
which is why i implemented a shadow

2468
01:29:51,600 --> 01:29:56,480
so you can see it shows under the player

2469
01:29:54,320 --> 01:29:58,239
to add a sense of immersion and not

2470
01:29:56,480 --> 01:30:00,880
break the suspension to disbelieve a

2471
01:29:58,239 --> 01:30:02,639
player has when playing a game

2472
01:30:00,880 --> 01:30:04,480
which is obviously quite useful for my

2473
01:30:02,639 --> 01:30:05,760
purposes

2474
01:30:04,480 --> 01:30:08,080
but

2475
01:30:05,760 --> 01:30:10,000
another success and

2476
01:30:08,080 --> 01:30:13,199
arguably another implementation of this

2477
01:30:10,000 --> 01:30:13,199
model can be seen here

2478
01:30:14,080 --> 01:30:18,159
now these shoot among you may notice

2479
01:30:16,080 --> 01:30:19,280
that there is a slight problem with this

2480
01:30:18,159 --> 01:30:20,719
clip

2481
01:30:19,280 --> 01:30:23,600
and that is that

2482
01:30:20,719 --> 01:30:25,679
it actually slightly overlaps in

2483
01:30:23,600 --> 01:30:28,239
a few of the frames

2484
01:30:25,679 --> 01:30:30,480
um which is to say that the shapes don't

2485
01:30:28,239 --> 01:30:32,800
draw in the right order now thinking

2486
01:30:30,480 --> 01:30:35,040
about draw or ordering

2487
01:30:32,800 --> 01:30:38,800
this is the system i came up with to do

2488
01:30:35,040 --> 01:30:41,440
it in isometric if you think uh you as

2489
01:30:38,800 --> 01:30:43,280
the audience looking at me and this pen

2490
01:30:41,440 --> 01:30:45,199
uh if you want to draw those to render

2491
01:30:43,280 --> 01:30:47,040
in the correct order if i'm in the back

2492
01:30:45,199 --> 01:30:50,480
you have to draw me first in a second so

2493
01:30:47,040 --> 01:30:51,360
this shows over me rather than behind me

2494
01:30:50,480 --> 01:30:52,880
and

2495
01:30:51,360 --> 01:30:54,560
in order to do this

2496
01:30:52,880 --> 01:30:56,000
i think of having a line that goes

2497
01:30:54,560 --> 01:30:57,440
straight from your vision straight

2498
01:30:56,000 --> 01:30:59,679
forward you don't have to care about

2499
01:30:57,440 --> 01:31:00,639
left or right or up or down

2500
01:30:59,679 --> 01:31:04,000
um

2501
01:31:00,639 --> 01:31:06,080
but if you look at me and then this pen

2502
01:31:04,000 --> 01:31:08,320
you think about this pen is actually

2503
01:31:06,080 --> 01:31:10,800
closer on that line going straight to

2504
01:31:08,320 --> 01:31:12,639
you than i am which is why this pen is

2505
01:31:10,800 --> 01:31:13,840
closer to you and should be rendered

2506
01:31:12,639 --> 01:31:15,840
last

2507
01:31:13,840 --> 01:31:17,600
and that's essentially what i did by

2508
01:31:15,840 --> 01:31:20,800
making a function that gives you the

2509
01:31:17,600 --> 01:31:22,800
normal to the to the uh camera for the

2510
01:31:20,800 --> 01:31:25,920
isometrics in 3d or for the

2511
01:31:22,800 --> 01:31:27,520
mathematically uninitiated of you and i

2512
01:31:25,920 --> 01:31:28,960
hope you look forward to that math i

2513
01:31:27,520 --> 01:31:30,960
certainly did

2514
01:31:28,960 --> 01:31:32,719
effectively just a line that goes

2515
01:31:30,960 --> 01:31:34,159
straight out from you like i was talking

2516
01:31:32,719 --> 01:31:38,239
about

2517
01:31:34,159 --> 01:31:40,159
i was able to sort shapes and points

2518
01:31:38,239 --> 01:31:41,360
to render in the correct order

2519
01:31:40,159 --> 01:31:43,679
however

2520
01:31:41,360 --> 01:31:45,920
in saying shapes it only really works

2521
01:31:43,679 --> 01:31:48,560
for points you can see in this example

2522
01:31:45,920 --> 01:31:49,840
here that this terrain that's generated

2523
01:31:48,560 --> 01:31:51,920
you can

2524
01:31:49,840 --> 01:31:53,520
see it drawn in the correct aura order

2525
01:31:51,920 --> 01:31:55,199
as the mountain moves closer to the

2526
01:31:53,520 --> 01:31:57,840
camera everything behind it is not

2527
01:31:55,199 --> 01:32:00,159
visible to you

2528
01:31:57,840 --> 01:32:02,000
that's exactly as you'd expect but then

2529
01:32:00,159 --> 01:32:03,600
why with those rectangular prisms does

2530
01:32:02,000 --> 01:32:06,719
it not always work

2531
01:32:03,600 --> 01:32:09,600
and as i understand it that's simply

2532
01:32:06,719 --> 01:32:12,480
a problem with discrete mathematics see

2533
01:32:09,600 --> 01:32:14,719
my formula is just built for working out

2534
01:32:12,480 --> 01:32:16,480
the position that it has in that line

2535
01:32:14,719 --> 01:32:18,239
which is only a point

2536
01:32:16,480 --> 01:32:19,920
not something that has width and height

2537
01:32:18,239 --> 01:32:21,040
and depth

2538
01:32:19,920 --> 01:32:22,560
and that's definitely something i'm

2539
01:32:21,040 --> 01:32:24,560
going to be working on in the future to

2540
01:32:22,560 --> 01:32:25,280
try and overcome that or work way around

2541
01:32:24,560 --> 01:32:27,600
it

2542
01:32:25,280 --> 01:32:30,239
but enough of these limitations

2543
01:32:27,600 --> 01:32:32,639
here's something i actually made with it

2544
01:32:30,239 --> 01:32:34,560
what you see here on the screen is my

2545
01:32:32,639 --> 01:32:36,239
year 12 major work for software design

2546
01:32:34,560 --> 01:32:38,880
and development

2547
01:32:36,239 --> 01:32:41,280
this is called ecosim as it is a

2548
01:32:38,880 --> 01:32:44,719
simulator of ecosystems you have plants

2549
01:32:41,280 --> 01:32:46,239
and animals interacting eating drinking

2550
01:32:44,719 --> 01:32:48,000
all that stuff

2551
01:32:46,239 --> 01:32:51,520
and the best way to think of to deliver

2552
01:32:48,000 --> 01:32:53,760
that was an isometric 3d to give it 2d

2553
01:32:51,520 --> 01:32:55,440
would mean have to flatten out animals

2554
01:32:53,760 --> 01:32:57,840
or have them very represented by a

2555
01:32:55,440 --> 01:33:00,000
character and while that works i do

2556
01:32:57,840 --> 01:33:02,960
value the immersion even if it is only

2557
01:33:00,000 --> 01:33:03,679
the emotion offered by a voxelized model

2558
01:33:02,960 --> 01:33:07,520
like

2559
01:33:03,679 --> 01:33:07,520
minecraft or terrarium for example

2560
01:33:07,600 --> 01:33:11,840
but in doing this you can rotate the

2561
01:33:09,520 --> 01:33:13,280
models in 3d you can see them moving

2562
01:33:11,840 --> 01:33:14,719
around and even where they're going to

2563
01:33:13,280 --> 01:33:16,000
move as they point in that direction

2564
01:33:14,719 --> 01:33:18,560
when they walk

2565
01:33:16,000 --> 01:33:20,800
and you can notice the view ordering is

2566
01:33:18,560 --> 01:33:22,800
almost correct as when something walks

2567
01:33:20,800 --> 01:33:24,719
behind the tree it does actually go

2568
01:33:22,800 --> 01:33:26,080
behind it and in front of it when it's

2569
01:33:24,719 --> 01:33:28,880
in front

2570
01:33:26,080 --> 01:33:31,360
which is definitely a success

2571
01:33:28,880 --> 01:33:33,600
however in saying that this did require

2572
01:33:31,360 --> 01:33:35,280
significant optimization

2573
01:33:33,600 --> 01:33:36,960
now for any of you that have tried to

2574
01:33:35,280 --> 01:33:39,520
make a rather significant project in

2575
01:33:36,960 --> 01:33:41,920
python probably well-known does not

2576
01:33:39,520 --> 01:33:44,719
always run the fastest

2577
01:33:41,920 --> 01:33:46,880
and a lot of that is definitely on me i

2578
01:33:44,719 --> 01:33:48,080
am not by any means the most experienced

2579
01:33:46,880 --> 01:33:49,199
programmer

2580
01:33:48,080 --> 01:33:51,280
um

2581
01:33:49,199 --> 01:33:53,760
and what i ended up coming up with was

2582
01:33:51,280 --> 01:33:56,639
two ways to optimize it the first way is

2583
01:33:53,760 --> 01:33:58,400
thinking of um as you see on the screen

2584
01:33:56,639 --> 01:34:00,000
here there are only certain tiles that

2585
01:33:58,400 --> 01:34:01,600
are actually visible to you you can

2586
01:34:00,000 --> 01:34:03,760
imagine that whole bunch of files are

2587
01:34:01,600 --> 01:34:05,280
off screen that aren't and drawing those

2588
01:34:03,760 --> 01:34:07,840
is a waste of processing power

2589
01:34:05,280 --> 01:34:10,639
especially if you are trying to do that

2590
01:34:07,840 --> 01:34:13,520
60 times a second

2591
01:34:10,639 --> 01:34:17,360
um so i i have done that by thinking of

2592
01:34:13,520 --> 01:34:18,800
each row as a line and doing some linear

2593
01:34:17,360 --> 01:34:20,800
but you can work out when the line

2594
01:34:18,800 --> 01:34:22,159
travels off the screen and that's the

2595
01:34:20,800 --> 01:34:24,320
wellington

2596
01:34:22,159 --> 01:34:26,080
but what if you were to zoom out you

2597
01:34:24,320 --> 01:34:28,320
would still be seeing everything and

2598
01:34:26,080 --> 01:34:30,719
optimization wouldn't help you there

2599
01:34:28,320 --> 01:34:32,639
because there's nothing off screen

2600
01:34:30,719 --> 01:34:35,920
so the next idea i had was just to

2601
01:34:32,639 --> 01:34:38,159
pre-render the sprite so turn a 3d map

2602
01:34:35,920 --> 01:34:40,239
into a 2d sprite which you can draw on

2603
01:34:38,159 --> 01:34:42,880
the screen each frame a lot more easily

2604
01:34:40,239 --> 01:34:44,400
than you can render every single tile

2605
01:34:42,880 --> 01:34:46,080
and that's what i ended up going with so

2606
01:34:44,400 --> 01:34:49,440
in this game there is a loading screen

2607
01:34:46,080 --> 01:34:51,920
as you open it up to pre-referendum

2608
01:34:49,440 --> 01:34:53,119
but moving into the future what i really

2609
01:34:51,920 --> 01:34:54,960
hope to do

2610
01:34:53,119 --> 01:34:56,239
is

2611
01:34:54,960 --> 01:34:57,840
to

2612
01:34:56,239 --> 01:35:00,320
kind of chunk these pre-rendered

2613
01:34:57,840 --> 01:35:03,360
renderings so to pre-render every eight

2614
01:35:00,320 --> 01:35:05,360
by eight for example tile map and then

2615
01:35:03,360 --> 01:35:08,560
uh

2616
01:35:05,360 --> 01:35:10,560
draw that as a texture-mapped

2617
01:35:08,560 --> 01:35:12,320
block of its own

2618
01:35:10,560 --> 01:35:14,080
which i think is an achievable goal for

2619
01:35:12,320 --> 01:35:16,880
the future

2620
01:35:14,080 --> 01:35:19,119
although it may take me years if i am to

2621
01:35:16,880 --> 01:35:20,560
do it all myself but enough about all

2622
01:35:19,119 --> 01:35:22,320
that

2623
01:35:20,560 --> 01:35:24,320
for those among you that are interested

2624
01:35:22,320 --> 01:35:26,320
as i know i certainly would be

2625
01:35:24,320 --> 01:35:29,280
this is the mathematics for this this

2626
01:35:26,320 --> 01:35:30,639
transformation of turning um the real

2627
01:35:29,280 --> 01:35:32,320
in-game coordinates into drawing

2628
01:35:30,639 --> 01:35:34,000
coordinates as you can see on the screen

2629
01:35:32,320 --> 01:35:35,199
here

2630
01:35:34,000 --> 01:35:37,840
what you have

2631
01:35:35,199 --> 01:35:39,600
is first off you have your decode as

2632
01:35:37,840 --> 01:35:41,840
your drawing coordinates your arco as

2633
01:35:39,600 --> 01:35:44,000
your real or in-game coordinates add for

2634
01:35:41,840 --> 01:35:46,800
the camera offset or add

2635
01:35:44,000 --> 01:35:49,360
as in addition uh zoom for how the

2636
01:35:46,800 --> 01:35:51,360
zoomed in the camera is

2637
01:35:49,360 --> 01:35:52,960
and then what you have is the real

2638
01:35:51,360 --> 01:35:55,119
coordinate

2639
01:35:52,960 --> 01:35:57,119
time the real x coordinate times sine of

2640
01:35:55,119 --> 01:35:58,719
theta plus the real y order times cos of

2641
01:35:57,119 --> 01:36:01,840
theta and you're talking about your

2642
01:35:58,719 --> 01:36:03,840
horizontal drawn coordinate and then for

2643
01:36:01,840 --> 01:36:05,119
the horizontal vertical component you

2644
01:36:03,840 --> 01:36:06,159
have

2645
01:36:05,119 --> 01:36:08,239
the real

2646
01:36:06,159 --> 01:36:11,920
in game x times cos of theta minus the

2647
01:36:08,239 --> 01:36:13,600
real y times sine of theta and you may

2648
01:36:11,920 --> 01:36:15,840
be wondering why i have a different sine

2649
01:36:13,600 --> 01:36:16,960
theta and closely on different lines and

2650
01:36:15,840 --> 01:36:18,480
to put it

2651
01:36:16,960 --> 01:36:20,320
moral assembly

2652
01:36:18,480 --> 01:36:22,159
that's basically a 90 degree rotation

2653
01:36:20,320 --> 01:36:24,400
because if you're looking straight down

2654
01:36:22,159 --> 01:36:25,679
a square is always supposed to have a

2655
01:36:24,400 --> 01:36:27,360
right angle

2656
01:36:25,679 --> 01:36:30,080
and i wanted to guarantee you that but

2657
01:36:27,360 --> 01:36:32,639
you might also be thinking well

2658
01:36:30,080 --> 01:36:34,719
what's this other cause and sign

2659
01:36:32,639 --> 01:36:35,760
and that

2660
01:36:34,719 --> 01:36:37,920
is

2661
01:36:35,760 --> 01:36:39,760
the vertical angle so if you're looking

2662
01:36:37,920 --> 01:36:43,040
at something from the side something

2663
01:36:39,760 --> 01:36:44,560
horizontal as you get closer to 90

2664
01:36:43,040 --> 01:36:47,119
degrees you don't see it at all it

2665
01:36:44,560 --> 01:36:49,520
becomes invisible it goes to zero much

2666
01:36:47,119 --> 01:36:52,080
like how cos of any angle goes to zero

2667
01:36:49,520 --> 01:36:55,440
at ninety degrees and similarly for your

2668
01:36:52,080 --> 01:36:56,800
real z times sine of alpha

2669
01:36:55,440 --> 01:36:58,480
it's much the same as you look from

2670
01:36:56,800 --> 01:37:01,440
directly above

2671
01:36:58,480 --> 01:37:05,040
well as you approach zero

2672
01:37:01,440 --> 01:37:07,679
goes to zero which is why you have sine

2673
01:37:05,040 --> 01:37:09,360
so thank you very much all for listening

2674
01:37:07,679 --> 01:37:11,119
i'll be in the comments if you have any

2675
01:37:09,360 --> 01:37:12,960
questions and thank you all to my family

2676
01:37:11,119 --> 01:37:13,760
my amazing teachers miss hogan and mr

2677
01:37:12,960 --> 01:37:16,719
till

2678
01:37:13,760 --> 01:37:17,679
and i'll see you all soon

2679
01:37:16,719 --> 01:37:19,679
thank you

2680
01:37:17,679 --> 01:37:21,679
very much for listening and have a great

2681
01:37:19,679 --> 01:37:23,600
python

2682
01:37:21,679 --> 01:37:25,920
see you around

2683
01:37:23,600 --> 01:37:28,719
fantastic thanks so much

2684
01:37:25,920 --> 01:37:30,639
now bella is our next speaker and she is

2685
01:37:28,719 --> 01:37:32,880
a repeat offender on the student

2686
01:37:30,639 --> 01:37:35,360
showcase lineup this year we'll be

2687
01:37:32,880 --> 01:37:37,440
hearing about tweet aware using ai to

2688
01:37:35,360 --> 01:37:38,960
deconstruct your twitter timeline

2689
01:37:37,440 --> 01:37:40,880
when i asked bella how long she'd been

2690
01:37:38,960 --> 01:37:43,840
programming she replied nine years in

2691
01:37:40,880 --> 01:37:46,239
total but six years properly

2692
01:37:43,840 --> 01:37:48,560
i'm not sure what designates proper

2693
01:37:46,239 --> 01:37:51,840
programming i assume from improper

2694
01:37:48,560 --> 01:37:53,280
programming but maybe i'll learn one day

2695
01:37:51,840 --> 01:37:56,159
when i asked bella what her favorite

2696
01:37:53,280 --> 01:37:58,560
emoji is she said heart's face which is

2697
01:37:56,159 --> 01:38:00,080
definitely something i can get behind

2698
01:37:58,560 --> 01:38:03,119
when i asked what she's learned recently

2699
01:38:00,080 --> 01:38:05,679
about computing she replied that almost

2700
01:38:03,119 --> 01:38:08,880
90 percent of the world's currency only

2701
01:38:05,679 --> 01:38:11,280
exists on computers so our physical cash

2702
01:38:08,880 --> 01:38:13,199
reserves are only 10

2703
01:38:11,280 --> 01:38:15,199
that's really interesting and i'm really

2704
01:38:13,199 --> 01:38:17,360
interested to hear her talk

2705
01:38:15,199 --> 01:38:18,880
hi my name's bella ryan i'm a year 12

2706
01:38:17,360 --> 01:38:20,560
student at roseville college i'm here

2707
01:38:18,880 --> 01:38:22,239
today to talk about my software design

2708
01:38:20,560 --> 01:38:24,880
development major project

2709
01:38:22,239 --> 01:38:26,960
tweet aware which is a python program

2710
01:38:24,880 --> 01:38:29,280
that uses the twitter

2711
01:38:26,960 --> 01:38:31,119
api to conduct sentiment analysis

2712
01:38:29,280 --> 01:38:34,000
funding users twitter feed

2713
01:38:31,119 --> 01:38:36,880
a bit of background to the project the

2714
01:38:34,000 --> 01:38:39,840
whole idea of this project was to find

2715
01:38:36,880 --> 01:38:41,679
a system either manual or automated

2716
01:38:39,840 --> 01:38:45,840
and find a way that you can

2717
01:38:41,679 --> 01:38:47,760
significantly improve the system so

2718
01:38:45,840 --> 01:38:49,360
when it comes to twitter it's not that i

2719
01:38:47,760 --> 01:38:51,520
necessarily found a problem with it it's

2720
01:38:49,360 --> 01:38:54,080
more like an opportunity as we

2721
01:38:51,520 --> 01:38:56,159
increasingly use social media

2722
01:38:54,080 --> 01:38:58,080
to connect with people as it becomes

2723
01:38:56,159 --> 01:39:00,239
more integral in our lives in terms of

2724
01:38:58,080 --> 01:39:03,360
education as well reflection and

2725
01:39:00,239 --> 01:39:05,920
analysis of what we are taking on

2726
01:39:03,360 --> 01:39:10,000
using these platforms becomes more and

2727
01:39:05,920 --> 01:39:11,520
more important uh in 2021 there are 192

2728
01:39:10,000 --> 01:39:13,840
million people actively using twitter in

2729
01:39:11,520 --> 01:39:15,440
the world and out of that user base 500

2730
01:39:13,840 --> 01:39:17,080
million tweets are posted every day

2731
01:39:15,440 --> 01:39:19,920
which is approximately

2732
01:39:17,080 --> 01:39:21,119
5787 tweets per second

2733
01:39:19,920 --> 01:39:22,800
so

2734
01:39:21,119 --> 01:39:24,560
the whole idea is that for each one of

2735
01:39:22,800 --> 01:39:26,400
those tweets there's at least one person

2736
01:39:24,560 --> 01:39:28,480
who's taking on the information

2737
01:39:26,400 --> 01:39:30,000
so how can i

2738
01:39:28,480 --> 01:39:30,719
improve twitter

2739
01:39:30,000 --> 01:39:33,280
to

2740
01:39:30,719 --> 01:39:34,719
give you a better idea of all right this

2741
01:39:33,280 --> 01:39:37,760
is what i'm taking on in terms of the

2742
01:39:34,719 --> 01:39:40,000
idea but reflect on whether

2743
01:39:37,760 --> 01:39:41,679
the way that that idea is put across is

2744
01:39:40,000 --> 01:39:43,679
positive negative or neutral so that you

2745
01:39:41,679 --> 01:39:45,840
can further reflect on how that might

2746
01:39:43,679 --> 01:39:48,639
impact

2747
01:39:45,840 --> 01:39:49,600
other aspects of the information

2748
01:39:48,639 --> 01:39:51,760
so

2749
01:39:49,600 --> 01:39:54,560
first off i started just using the web

2750
01:39:51,760 --> 01:39:57,360
documentation between development uh and

2751
01:39:54,560 --> 01:39:58,800
i found a get request and it was a get

2752
01:39:57,360 --> 01:40:00,320
request for user's timeline i was super

2753
01:39:58,800 --> 01:40:02,320
keen i was like yes this is exactly what

2754
01:40:00,320 --> 01:40:04,639
i'm looking for ran it through with all

2755
01:40:02,320 --> 01:40:06,639
my bearer authentication all of that

2756
01:40:04,639 --> 01:40:09,119
and i got this i was like sweet tweet

2757
01:40:06,639 --> 01:40:11,119
text it's working this is amazing

2758
01:40:09,119 --> 01:40:12,480
uh then when i went across to the

2759
01:40:11,119 --> 01:40:14,480
account that i used it on which was

2760
01:40:12,480 --> 01:40:16,239
twitterdev i realized that what i was

2761
01:40:14,480 --> 01:40:17,760
actually getting was

2762
01:40:16,239 --> 01:40:18,800
all of their tweets read tweets and

2763
01:40:17,760 --> 01:40:21,440
comments

2764
01:40:18,800 --> 01:40:23,840
so twitter the twitter api doesn't

2765
01:40:21,440 --> 01:40:25,600
actually support you

2766
01:40:23,840 --> 01:40:27,840
just putting in someone's handle and

2767
01:40:25,600 --> 01:40:30,239
then being able to get the feed as they

2768
01:40:27,840 --> 01:40:33,040
see it so i had to recreate

2769
01:40:30,239 --> 01:40:35,119
it and instead model the feed

2770
01:40:33,040 --> 01:40:37,119
uh this is one i used onto postman which

2771
01:40:35,119 --> 01:40:38,800
is an absolutely fabulous

2772
01:40:37,119 --> 01:40:40,560
program to use if you're trying to

2773
01:40:38,800 --> 01:40:42,639
figure out how to use an api especially

2774
01:40:40,560 --> 01:40:44,639
if the people who developed the api have

2775
01:40:42,639 --> 01:40:47,679
created a dashboard like this

2776
01:40:44,639 --> 01:40:50,239
um using that i was able to figure out

2777
01:40:47,679 --> 01:40:52,239
the order of apis i'd need to use to

2778
01:40:50,239 --> 01:40:53,760
recreate someone's feed

2779
01:40:52,239 --> 01:40:55,280
so to do this

2780
01:40:53,760 --> 01:40:57,679
i take on

2781
01:40:55,280 --> 01:40:59,440
someone's i take in input from the user

2782
01:40:57,679 --> 01:41:01,920
which is their handle which i then send

2783
01:40:59,440 --> 01:41:03,679
to the api and i get back their user id

2784
01:41:01,920 --> 01:41:05,760
which i then send to the api again to

2785
01:41:03,679 --> 01:41:06,639
get a list of every um user that they

2786
01:41:05,760 --> 01:41:07,600
follow

2787
01:41:06,639 --> 01:41:09,920
i

2788
01:41:07,600 --> 01:41:11,840
send a request for every single user on

2789
01:41:09,920 --> 01:41:14,320
that list to get all their ids and then

2790
01:41:11,840 --> 01:41:15,360
i use the timeline request that i showed

2791
01:41:14,320 --> 01:41:16,159
earlier

2792
01:41:15,360 --> 01:41:19,440
to

2793
01:41:16,159 --> 01:41:21,679
get their um five most recent tweets

2794
01:41:19,440 --> 01:41:24,880
from their account obviously that poses

2795
01:41:21,679 --> 01:41:26,639
floor floors with the model of how i'm

2796
01:41:24,880 --> 01:41:28,000
recreating people's feeds because if

2797
01:41:26,639 --> 01:41:29,360
you've got someone who does two tweets

2798
01:41:28,000 --> 01:41:31,040
in their lifetime versus say a

2799
01:41:29,360 --> 01:41:33,199
journalist who does two tweets every

2800
01:41:31,040 --> 01:41:34,239
five minutes uh

2801
01:41:33,199 --> 01:41:37,040
their

2802
01:41:34,239 --> 01:41:39,679
representation in the data

2803
01:41:37,040 --> 01:41:42,000
is not is skewed in a way that's not

2804
01:41:39,679 --> 01:41:46,159
represented representative of an actual

2805
01:41:42,000 --> 01:41:47,679
feed which is why it's a model um i then

2806
01:41:46,159 --> 01:41:48,880
send it through

2807
01:41:47,679 --> 01:41:51,040
a

2808
01:41:48,880 --> 01:41:54,159
scrubbing program that takes out things

2809
01:41:51,040 --> 01:41:56,159
like new lines and other html css

2810
01:41:54,159 --> 01:41:58,960
elements that sometimes come through in

2811
01:41:56,159 --> 01:42:01,119
the tweet text that are fantastic for

2812
01:41:58,960 --> 01:42:03,360
displaying it on twitter but not very

2813
01:42:01,119 --> 01:42:06,159
useful for the purpose

2814
01:42:03,360 --> 01:42:07,760
of sentiment analysis

2815
01:42:06,159 --> 01:42:09,600
next i send it through the sentiment

2816
01:42:07,760 --> 01:42:11,840
analysis algorithm

2817
01:42:09,600 --> 01:42:15,920
for this i use text blob

2818
01:42:11,840 --> 01:42:17,280
and so i have a list of all of the um

2819
01:42:15,920 --> 01:42:19,920
clean tweets

2820
01:42:17,280 --> 01:42:22,560
and i send each tweet through

2821
01:42:19,920 --> 01:42:24,560
text blog which then i use the sentiment

2822
01:42:22,560 --> 01:42:25,520
polarity function

2823
01:42:24,560 --> 01:42:27,679
to

2824
01:42:25,520 --> 01:42:29,600
um give them h a value of either

2825
01:42:27,679 --> 01:42:31,679
positive mutual or negative

2826
01:42:29,600 --> 01:42:33,199
and then when they're outputted from

2827
01:42:31,679 --> 01:42:35,440
that function

2828
01:42:33,199 --> 01:42:37,760
uh they're all input into a dictionary

2829
01:42:35,440 --> 01:42:40,000
where the key is the tweet itself and

2830
01:42:37,760 --> 01:42:43,199
then the value is the

2831
01:42:40,000 --> 01:42:44,080
sentiment of positive mutual or negative

2832
01:42:43,199 --> 01:42:47,119
um

2833
01:42:44,080 --> 01:42:49,520
next is output so i have three kinds of

2834
01:42:47,119 --> 01:42:54,159
output for this project i had a

2835
01:42:49,520 --> 01:42:56,000
txt file that had the percentages

2836
01:42:54,159 --> 01:42:58,800
of positive negative and neutral within

2837
01:42:56,000 --> 01:43:01,760
the data set i had a

2838
01:42:58,800 --> 01:43:03,600
png which is the pie graph that you see

2839
01:43:01,760 --> 01:43:04,800
there of again positive negative and

2840
01:43:03,600 --> 01:43:08,840
neutral

2841
01:43:04,800 --> 01:43:12,800
and then i also had output within

2842
01:43:08,840 --> 01:43:13,920
the command line interface which was the

2843
01:43:12,800 --> 01:43:15,600
five

2844
01:43:13,920 --> 01:43:17,199
the first five in each list of positive

2845
01:43:15,600 --> 01:43:20,560
and negative tweets

2846
01:43:17,199 --> 01:43:23,600
um for the user to look at each time

2847
01:43:20,560 --> 01:43:24,400
they run the program for each handle

2848
01:43:23,600 --> 01:43:27,840
uh

2849
01:43:24,400 --> 01:43:29,600
from this you i decided to

2850
01:43:27,840 --> 01:43:30,480
run again with a real world example of

2851
01:43:29,600 --> 01:43:33,119
how

2852
01:43:30,480 --> 01:43:35,840
this program could be taken a little bit

2853
01:43:33,119 --> 01:43:38,239
further so as we're in lockdown now

2854
01:43:35,840 --> 01:43:40,320
a lot of us are receiving news through

2855
01:43:38,239 --> 01:43:42,159
social media whether

2856
01:43:40,320 --> 01:43:44,800
it's the um

2857
01:43:42,159 --> 01:43:46,560
11am livestream on facebook

2858
01:43:44,800 --> 01:43:48,239
or it's the little

2859
01:43:46,560 --> 01:43:49,840
uh little bits that have had through

2860
01:43:48,239 --> 01:43:53,360
instagram or twitter in forms of

2861
01:43:49,840 --> 01:43:55,760
infographics and so on so using my

2862
01:43:53,360 --> 01:43:58,560
testing twitter account i followed a

2863
01:43:55,760 --> 01:44:03,040
bunch of australian

2864
01:43:58,560 --> 01:44:05,920
news platforms and then ran them through

2865
01:44:03,040 --> 01:44:08,080
and got the outfit of five most recent

2866
01:44:05,920 --> 01:44:10,800
positive tweets negative tweets tweet

2867
01:44:08,080 --> 01:44:12,560
percentages and then the paragraph

2868
01:44:10,800 --> 01:44:13,360
um

2869
01:44:12,560 --> 01:44:15,199
which

2870
01:44:13,360 --> 01:44:16,719
i think reflects

2871
01:44:15,199 --> 01:44:17,920
the next step before i want this project

2872
01:44:16,719 --> 01:44:18,840
to go

2873
01:44:17,920 --> 01:44:21,920
which

2874
01:44:18,840 --> 01:44:21,920
is uh

2875
01:44:22,159 --> 01:44:25,600
more

2876
01:44:23,040 --> 01:44:28,639
more kind of data visualization so word

2877
01:44:25,600 --> 01:44:30,880
clouds um

2878
01:44:28,639 --> 01:44:34,719
and things like that looking at

2879
01:44:30,880 --> 01:44:36,239
data historically so month to month what

2880
01:44:34,719 --> 01:44:38,880
sentiment scores you're getting through

2881
01:44:36,239 --> 01:44:40,880
your feed and also going back to the

2882
01:44:38,880 --> 01:44:42,239
original plan for my project which was

2883
01:44:40,880 --> 01:44:44,480
to have

2884
01:44:42,239 --> 01:44:47,040
live sentiment analysis on the feed

2885
01:44:44,480 --> 01:44:48,800
itself i think i could achieve that that

2886
01:44:47,040 --> 01:44:49,679
would be really awesome

2887
01:44:48,800 --> 01:44:51,600
uh

2888
01:44:49,679 --> 01:44:53,440
yeah so thank you for listening to my

2889
01:44:51,600 --> 01:44:55,679
presentation i guess the main takeaway

2890
01:44:53,440 --> 01:44:57,760
that i want from this is when you're

2891
01:44:55,679 --> 01:44:59,440
given the chance to do a project in

2892
01:44:57,760 --> 01:45:00,639
school pick something that you love and

2893
01:44:59,440 --> 01:45:02,400
pick something that you can keep on

2894
01:45:00,639 --> 01:45:04,320
working on because

2895
01:45:02,400 --> 01:45:06,880
during the hard knock times of the

2896
01:45:04,320 --> 01:45:08,880
project itself it kind of prompts you to

2897
01:45:06,880 --> 01:45:10,719
actually continue working on the project

2898
01:45:08,880 --> 01:45:12,400
but afterwards it gives you a fantastic

2899
01:45:10,719 --> 01:45:13,440
springboard to jump off into your next

2900
01:45:12,400 --> 01:45:15,040
project

2901
01:45:13,440 --> 01:45:17,520
thank you so much

2902
01:45:15,040 --> 01:45:19,600
now our final speaker today is menindra

2903
01:45:17,520 --> 01:45:21,840
who'll be talking about de-blurring with

2904
01:45:19,600 --> 01:45:22,960
deep learning which is perhaps more of a

2905
01:45:21,840 --> 01:45:24,800
tongue twister than i should have

2906
01:45:22,960 --> 01:45:26,719
finished today on

2907
01:45:24,800 --> 01:45:29,440
menindra has been programming for three

2908
01:45:26,719 --> 01:45:31,760
years and his favorite emoji is the dead

2909
01:45:29,440 --> 01:45:34,320
face emoji specifically he said he

2910
01:45:31,760 --> 01:45:36,320
thinks it's quite funny

2911
01:45:34,320 --> 01:45:37,679
when i asked what recent topic of

2912
01:45:36,320 --> 01:45:40,159
computer science he's been learning

2913
01:45:37,679 --> 01:45:42,080
about he said machine learning deep

2914
01:45:40,159 --> 01:45:44,639
learning and specifically supervised

2915
01:45:42,080 --> 01:45:47,679
learning with neural networks and i'm

2916
01:45:44,639 --> 01:45:50,400
really excited to hear his talk

2917
01:45:47,679 --> 01:45:52,080
hello pycon my name is melinda de mel

2918
01:45:50,400 --> 01:45:54,639
and i'd like to present my project d

2919
01:45:52,080 --> 01:45:56,080
blurring with deep learning

2920
01:45:54,639 --> 01:45:57,920
so i first began the year writing a

2921
01:45:56,080 --> 01:45:59,280
report on back propagation and the

2922
01:45:57,920 --> 01:46:01,199
report presents an intuitive and

2923
01:45:59,280 --> 01:46:03,920
comprehensive analysis on the algorithm

2924
01:46:01,199 --> 01:46:05,360
it was a lot of fun to write

2925
01:46:03,920 --> 01:46:06,480
so i then wanted to extend my newly

2926
01:46:05,360 --> 01:46:09,280
found knowledge under a couple of

2927
01:46:06,480 --> 01:46:11,440
project ideas so i thought of q learning

2928
01:46:09,280 --> 01:46:13,280
the genetic algorithm neat machine

2929
01:46:11,440 --> 01:46:14,400
learning with robotics quantum machine

2930
01:46:13,280 --> 01:46:16,400
learning

2931
01:46:14,400 --> 01:46:18,239
i thought of fluid simulation binary

2932
01:46:16,400 --> 01:46:20,000
systems

2933
01:46:18,239 --> 01:46:22,080
but then i realized that all of these

2934
01:46:20,000 --> 01:46:23,920
ideas were kind of irrelevant to bad

2935
01:46:22,080 --> 01:46:25,119
propagation and out of scope for my

2936
01:46:23,920 --> 01:46:27,360
project

2937
01:46:25,119 --> 01:46:28,880
so i then moved on to computer vision

2938
01:46:27,360 --> 01:46:30,239
and of computer vision i chose image

2939
01:46:28,880 --> 01:46:31,679
restoration

2940
01:46:30,239 --> 01:46:33,520
i chose it mainly because it was

2941
01:46:31,679 --> 01:46:34,800
relevant to back propagation and

2942
01:46:33,520 --> 01:46:36,480
actually initially i was going to

2943
01:46:34,800 --> 01:46:38,560
restore color into black and white

2944
01:46:36,480 --> 01:46:40,480
images but then i realized this is kind

2945
01:46:38,560 --> 01:46:43,040
of deprecated because no one takes like

2946
01:46:40,480 --> 01:46:44,480
on my photos these days so i then moved

2947
01:46:43,040 --> 01:46:46,239
on to deep blurring and i realized it's

2948
01:46:44,480 --> 01:46:48,400
more relevant because people do take

2949
01:46:46,239 --> 01:46:50,239
blurred photos these days and there

2950
01:46:48,400 --> 01:46:52,880
aren't many implementations or solutions

2951
01:46:50,239 --> 01:46:54,800
out there to resolve this

2952
01:46:52,880 --> 01:46:57,280
i'll now go through a brief overview of

2953
01:46:54,800 --> 01:46:59,199
how deep learning works so first we have

2954
01:46:57,280 --> 01:47:01,280
neural networks and neural networks are

2955
01:46:59,199 --> 01:47:02,719
modeled off of the human brain where we

2956
01:47:01,280 --> 01:47:04,639
can think of our brain as a bunch of

2957
01:47:02,719 --> 01:47:06,480
neurons uh connected to other neurons

2958
01:47:04,639 --> 01:47:08,639
through axioms

2959
01:47:06,480 --> 01:47:11,040
and similarly in neural networks we have

2960
01:47:08,639 --> 01:47:12,880
layers of neurons uh they're connected

2961
01:47:11,040 --> 01:47:14,560
to other layers of neurons with

2962
01:47:12,880 --> 01:47:16,719
weights where the blue ones are positive

2963
01:47:14,560 --> 01:47:19,920
and the red ones are negative and they

2964
01:47:16,719 --> 01:47:22,239
all affect each other accordingly

2965
01:47:19,920 --> 01:47:24,480
we also have biases which are associated

2966
01:47:22,239 --> 01:47:26,639
with each neuron and they just help with

2967
01:47:24,480 --> 01:47:28,159
the training process

2968
01:47:26,639 --> 01:47:30,400
so here's what a neuron might look like

2969
01:47:28,159 --> 01:47:33,760
we have the weights the bias the

2970
01:47:30,400 --> 01:47:35,040
activation and the way to calculate it

2971
01:47:33,760 --> 01:47:37,119
and here's what the network might look

2972
01:47:35,040 --> 01:47:39,520
like we have the layers

2973
01:47:37,119 --> 01:47:42,000
and the layer sizes and we also have a

2974
01:47:39,520 --> 01:47:46,000
way to generate the initial uh

2975
01:47:42,000 --> 01:47:46,000
rates connecting layers together

2976
01:47:46,480 --> 01:47:50,719
for propagation uh

2977
01:47:48,159 --> 01:47:53,840
for progression simply put it's just

2978
01:47:50,719 --> 01:47:56,159
taking an input and then applying the

2979
01:47:53,840 --> 01:47:58,480
weights and biases from each layer and

2980
01:47:56,159 --> 01:48:00,480
then passing it through the network uh

2981
01:47:58,480 --> 01:48:01,360
in real life this could be compared to

2982
01:48:00,480 --> 01:48:03,600
um

2983
01:48:01,360 --> 01:48:06,000
taking an input from our environment our

2984
01:48:03,600 --> 01:48:08,800
brain processing it and giving us an

2985
01:48:06,000 --> 01:48:10,639
up an output

2986
01:48:08,800 --> 01:48:12,400
so initially for a neural network this

2987
01:48:10,639 --> 01:48:14,480
causes high error between the expected

2988
01:48:12,400 --> 01:48:16,560
output and the network's output

2989
01:48:14,480 --> 01:48:19,520
but once it's been trained um

2990
01:48:16,560 --> 01:48:21,280
it's actually what's used to predict uh

2991
01:48:19,520 --> 01:48:23,679
non-blood images in my case or the

2992
01:48:21,280 --> 01:48:23,679
answer

2993
01:48:24,239 --> 01:48:27,679
uh for propagation

2994
01:48:25,920 --> 01:48:30,800
uh this is what it would look like in

2995
01:48:27,679 --> 01:48:32,320
code so we have um we go through each

2996
01:48:30,800 --> 01:48:35,119
layer and we go through each node in the

2997
01:48:32,320 --> 01:48:36,960
layer and we pass it the previous layer

2998
01:48:35,119 --> 01:48:39,280
and then we

2999
01:48:36,960 --> 01:48:40,639
calculate uh all the weights and all the

3000
01:48:39,280 --> 01:48:42,639
activations from the previous layer and

3001
01:48:40,639 --> 01:48:43,360
we find an activation for that node in

3002
01:48:42,639 --> 01:48:44,840
the

3003
01:48:43,360 --> 01:48:47,280
next

3004
01:48:44,840 --> 01:48:50,239
layer so this is what the network might

3005
01:48:47,280 --> 01:48:52,880
look like so we could uh take the input

3006
01:48:50,239 --> 01:48:54,159
uh we pass it to uh the processing layer

3007
01:48:52,880 --> 01:48:56,000
or the hidden layer and get an

3008
01:48:54,159 --> 01:48:57,760
activation and then from this we pass

3009
01:48:56,000 --> 01:48:59,199
through the next layer the output and we

3010
01:48:57,760 --> 01:49:01,600
get activations for all these

3011
01:48:59,199 --> 01:49:03,600
accordingly

3012
01:49:01,600 --> 01:49:05,679
uh back propagation so back propagation

3013
01:49:03,600 --> 01:49:07,199
really is the focus of my algorithm it's

3014
01:49:05,679 --> 01:49:09,440
like the

3015
01:49:07,199 --> 01:49:11,600
heart of the algorithm where without it

3016
01:49:09,440 --> 01:49:13,760
it won't be able to function uh the

3017
01:49:11,600 --> 01:49:15,679
structure of

3018
01:49:13,760 --> 01:49:17,040
the new network itself uh would

3019
01:49:15,679 --> 01:49:18,639
effectively be useless without back

3020
01:49:17,040 --> 01:49:20,080
propagation because you won't be able to

3021
01:49:18,639 --> 01:49:20,960
learn

3022
01:49:20,080 --> 01:49:22,880
so

3023
01:49:20,960 --> 01:49:24,560
here's the mathematics for it so

3024
01:49:22,880 --> 01:49:25,920
i don't want to read any of it but

3025
01:49:24,560 --> 01:49:27,199
this is for my report and i just want

3026
01:49:25,920 --> 01:49:30,080
you to acknowledge it was a lot of fun

3027
01:49:27,199 --> 01:49:31,760
to learn about read about and write

3028
01:49:30,080 --> 01:49:33,599
about and essentially what we're going

3029
01:49:31,760 --> 01:49:36,320
to take is this

3030
01:49:33,599 --> 01:49:37,360
derivative or

3031
01:49:36,320 --> 01:49:38,800
equation

3032
01:49:37,360 --> 01:49:40,320
per se so

3033
01:49:38,800 --> 01:49:43,119
you want to find the rate of change of

3034
01:49:40,320 --> 01:49:46,800
the cost with respect to each weight and

3035
01:49:43,119 --> 01:49:49,280
bias of the network so the cost is the

3036
01:49:46,800 --> 01:49:51,599
overall error of the network it's the

3037
01:49:49,280 --> 01:49:53,599
network's output and the expected output

3038
01:49:51,599 --> 01:49:56,400
and the difference and basically we want

3039
01:49:53,599 --> 01:49:58,320
to minimize this to make the network get

3040
01:49:56,400 --> 01:49:59,920
the expected output

3041
01:49:58,320 --> 01:50:02,159
and so effectively what do we apply the

3042
01:49:59,920 --> 01:50:04,080
chain rule to each meter as major aspect

3043
01:50:02,159 --> 01:50:05,360
of the

3044
01:50:04,080 --> 01:50:08,239
new network

3045
01:50:05,360 --> 01:50:08,239
and then we

3046
01:50:08,320 --> 01:50:13,040
find how each rate and bias

3047
01:50:10,560 --> 01:50:14,719
affect the overall cost and we can see

3048
01:50:13,040 --> 01:50:16,960
these how this is written in the code so

3049
01:50:14,719 --> 01:50:17,920
we have these highlighted boxes and you

3050
01:50:16,960 --> 01:50:19,599
can see

3051
01:50:17,920 --> 01:50:20,960
the yellow blocks we take the overall

3052
01:50:19,599 --> 01:50:23,920
cost respect to the output of the

3053
01:50:20,960 --> 01:50:26,000
network in the blue box we take the

3054
01:50:23,920 --> 01:50:27,599
output layer with respect to all the

3055
01:50:26,000 --> 01:50:30,320
previous layers

3056
01:50:27,599 --> 01:50:32,400
and then for each neuron in the previous

3057
01:50:30,320 --> 01:50:34,960
layers we take the uh neuron with

3058
01:50:32,400 --> 01:50:37,199
respect to each mate and bias and we get

3059
01:50:34,960 --> 01:50:39,679
the each uh each weight and biases error

3060
01:50:37,199 --> 01:50:40,800
on the overall cost

3061
01:50:39,679 --> 01:50:42,239
and so this is what it might look like

3062
01:50:40,800 --> 01:50:44,800
so we have

3063
01:50:42,239 --> 01:50:46,960
the input and then these numbers are the

3064
01:50:44,800 --> 01:50:48,639
effect on the overall cost so this

3065
01:50:46,960 --> 01:50:50,800
neuron here has negative effect on the

3066
01:50:48,639 --> 01:50:52,800
overall cost and then this neuron has a

3067
01:50:50,800 --> 01:50:54,080
positive effect on your overall cost and

3068
01:50:52,800 --> 01:50:56,320
basically you want to make these as

3069
01:50:54,080 --> 01:50:59,040
close as possible to zero so we get the

3070
01:50:56,320 --> 01:51:00,960
expected output

3071
01:50:59,040 --> 01:51:03,360
and so once we have these errors and

3072
01:51:00,960 --> 01:51:04,320
these costs for each neuron

3073
01:51:03,360 --> 01:51:05,760
we can

3074
01:51:04,320 --> 01:51:07,360
affect the

3075
01:51:05,760 --> 01:51:09,920
rates and biases accordingly by

3076
01:51:07,360 --> 01:51:12,639
increasing them up or down to reduce

3077
01:51:09,920 --> 01:51:12,639
this cost

3078
01:51:12,800 --> 01:51:16,560
and so once the network has been trained

3079
01:51:14,480 --> 01:51:17,760
we can then um

3080
01:51:16,560 --> 01:51:20,239
save the

3081
01:51:17,760 --> 01:51:22,480
the parameters or the weights and biases

3082
01:51:20,239 --> 01:51:24,560
and then we can use forward propagation

3083
01:51:22,480 --> 01:51:26,080
to simply pass an input and get a

3084
01:51:24,560 --> 01:51:27,760
predicted output which hopefully is

3085
01:51:26,080 --> 01:51:29,760
really accurate

3086
01:51:27,760 --> 01:51:31,360
so here's some training data so we have

3087
01:51:29,760 --> 01:51:34,400
here we have the error or the cost on

3088
01:51:31,360 --> 01:51:37,040
the y-axis and we have the iterations uh

3089
01:51:34,400 --> 01:51:38,400
the training process and we can see we

3090
01:51:37,040 --> 01:51:40,480
want to minimize this cost and it's

3091
01:51:38,400 --> 01:51:41,599
doing this but it's still relatively

3092
01:51:40,480 --> 01:51:44,239
high

3093
01:51:41,599 --> 01:51:46,400
and you can see in this one as well

3094
01:51:44,239 --> 01:51:49,360
it goes down but it's still like above

3095
01:51:46,400 --> 01:51:51,840
20 and this is quite bad and you can see

3096
01:51:49,360 --> 01:51:54,480
the result of this as well so

3097
01:51:51,840 --> 01:51:56,639
we have our blood image we have our

3098
01:51:54,480 --> 01:51:59,920
sharpen image and then we have

3099
01:51:56,639 --> 01:52:02,080
uh an output so this is a complete mess

3100
01:51:59,920 --> 01:52:03,199
it's completely unrecognizable from the

3101
01:52:02,080 --> 01:52:05,840
other two

3102
01:52:03,199 --> 01:52:07,119
and so i figured that maybe these images

3103
01:52:05,840 --> 01:52:09,520
were too big

3104
01:52:07,119 --> 01:52:10,960
their dimensions were like 500 by 637

3105
01:52:09,520 --> 01:52:12,639
pixels which is a lot of weights of

3106
01:52:10,960 --> 01:52:15,119
biases to compute

3107
01:52:12,639 --> 01:52:17,280
so i moved to a smaller data set and i

3108
01:52:15,119 --> 01:52:19,520
use the mnist the blurring set for

3109
01:52:17,280 --> 01:52:21,760
digits handwritten digits and we can see

3110
01:52:19,520 --> 01:52:24,000
a blurred six sharpen six

3111
01:52:21,760 --> 01:52:25,920
and the networks output which again

3112
01:52:24,000 --> 01:52:27,280
is still pretty bad but you can see that

3113
01:52:25,920 --> 01:52:30,560
like the bottom right half is being

3114
01:52:27,280 --> 01:52:32,320
sharpened which is kind of good and then

3115
01:52:30,560 --> 01:52:33,920
i realized that maybe this is due to the

3116
01:52:32,320 --> 01:52:37,199
training method so i changed from

3117
01:52:33,920 --> 01:52:39,280
stochastic going by each row

3118
01:52:37,199 --> 01:52:41,119
row by row sorry and then i just passed

3119
01:52:39,280 --> 01:52:43,760
the whole image to the network

3120
01:52:41,119 --> 01:52:46,639
and we saw that the

3121
01:52:43,760 --> 01:52:49,440
loss or the cost and the error decreased

3122
01:52:46,639 --> 01:52:51,920
very fast it went from like 200 to like

3123
01:52:49,440 --> 01:52:53,199
below 10 in like the first iteration and

3124
01:52:51,920 --> 01:52:54,639
it stayed there which is really really

3125
01:52:53,199 --> 01:52:56,480
good

3126
01:52:54,639 --> 01:52:58,080
and we can see we get some results from

3127
01:52:56,480 --> 01:53:00,480
this so we have a blurred five and a

3128
01:52:58,080 --> 01:53:02,480
sharpen five a couple of results i got

3129
01:53:00,480 --> 01:53:03,920
and these are very sharp you can see the

3130
01:53:02,480 --> 01:53:05,520
network really picked up on the edges

3131
01:53:03,920 --> 01:53:06,800
and the curves of the fives

3132
01:53:05,520 --> 01:53:08,000
you can see some more examples we have

3133
01:53:06,800 --> 01:53:10,880
blurred five

3134
01:53:08,000 --> 01:53:13,040
we have these um

3135
01:53:10,880 --> 01:53:14,719
some more fives that these are

3136
01:53:13,040 --> 01:53:16,639
some of the better ones i had especially

3137
01:53:14,719 --> 01:53:18,639
this one really got the shape of the fly

3138
01:53:16,639 --> 01:53:21,360
if not if it's actually better than the

3139
01:53:18,639 --> 01:53:24,639
answer in my opinion

3140
01:53:21,360 --> 01:53:26,639
and so now i have two demos to show my

3141
01:53:24,639 --> 01:53:28,719
project so the first demo is the

3142
01:53:26,639 --> 01:53:30,320
training process

3143
01:53:28,719 --> 01:53:32,639
so we have a video here

3144
01:53:30,320 --> 01:53:34,719
so we have the blurred seven we have the

3145
01:53:32,639 --> 01:53:36,480
sharpen seven and then we have uh the

3146
01:53:34,719 --> 01:53:38,480
network output down here and we have the

3147
01:53:36,480 --> 01:53:40,320
graph showing the error with respect to

3148
01:53:38,480 --> 01:53:42,960
each training iteration and we can see

3149
01:53:40,320 --> 01:53:45,360
as it goes it slowly

3150
01:53:42,960 --> 01:53:48,080
the error slowly decreases and slowly

3151
01:53:45,360 --> 01:53:50,000
the server uh the network finds

3152
01:53:48,080 --> 01:53:51,920
a really good seven after a couple

3153
01:53:50,000 --> 01:53:54,719
iterations

3154
01:53:51,920 --> 01:53:56,800
like that and we can see again blurred

3155
01:53:54,719 --> 01:53:58,320
sharpen and a really really good output

3156
01:53:56,800 --> 01:54:01,520
right here

3157
01:53:58,320 --> 01:54:03,840
and my second um demo is going to be in

3158
01:54:01,520 --> 01:54:05,360
a network output so we can see we have a

3159
01:54:03,840 --> 01:54:08,080
blurred four here a sharpened floor here

3160
01:54:05,360 --> 01:54:10,880
and if i just for propagate to a network

3161
01:54:08,080 --> 01:54:12,880
and pass it this blood for

3162
01:54:10,880 --> 01:54:15,360
just like so

3163
01:54:12,880 --> 01:54:16,560
and then pass it right here

3164
01:54:15,360 --> 01:54:17,520
you can see

3165
01:54:16,560 --> 01:54:19,840
boom

3166
01:54:17,520 --> 01:54:21,920
blood fall sharpen floor and then our

3167
01:54:19,840 --> 01:54:23,920
network output which again is really

3168
01:54:21,920 --> 01:54:25,119
sharp like it's almost it's actually

3169
01:54:23,920 --> 01:54:28,000
better than the

3170
01:54:25,119 --> 01:54:30,320
actual answer

3171
01:54:28,000 --> 01:54:32,560
and so now

3172
01:54:30,320 --> 01:54:32,560
um

3173
01:54:32,960 --> 01:54:38,560
we have that's our probe my project so

3174
01:54:36,639 --> 01:54:40,880
overall this project was a lot of fun to

3175
01:54:38,560 --> 01:54:42,239
write uh write the code for it was a lot

3176
01:54:40,880 --> 01:54:44,239
of fun to learn about it was a good

3177
01:54:42,239 --> 01:54:45,280
experience i had a lot of fun

3178
01:54:44,239 --> 01:54:46,560
um

3179
01:54:45,280 --> 01:54:47,679
and i recommend it to anyone out there

3180
01:54:46,560 --> 01:54:49,119
who wants to start machine learning

3181
01:54:47,679 --> 01:54:51,360
because this is my machine first machine

3182
01:54:49,119 --> 01:54:52,880
learning project and if anyone has any

3183
01:54:51,360 --> 01:54:55,040
questions about a project or how it

3184
01:54:52,880 --> 01:54:57,920
works i'll just leave it in the chat and

3185
01:54:55,040 --> 01:55:00,080
thank you for watching my presentation

3186
01:54:57,920 --> 01:55:02,239
thank you so much for joining me and

3187
01:55:00,080 --> 01:55:04,239
thank you again to our fantastic

3188
01:55:02,239 --> 01:55:06,639
speakers and i will pass it back to

3189
01:55:04,239 --> 01:55:08,560
future nikki

3190
01:55:06,639 --> 01:55:12,159
yeah absolutely

3191
01:55:08,560 --> 01:55:14,480
so uh please we want we want a really

3192
01:55:12,159 --> 01:55:17,040
big range from beginner to absolute

3193
01:55:14,480 --> 01:55:19,679
expert in student slam downs in the

3194
01:55:17,040 --> 01:55:21,599
future

3195
01:55:19,679 --> 01:55:23,840
we don't want to we want to encourage

3196
01:55:21,599 --> 01:55:26,639
people and build them up and this is a

3197
01:55:23,840 --> 01:55:30,159
fantastic way to do it yes and and thank

3198
01:55:26,639 --> 01:55:33,760
you all everyone who has been in the

3199
01:55:30,159 --> 01:55:36,480
um in the chat for showing so much um

3200
01:55:33,760 --> 01:55:39,920
love and support for these fantastic

3201
01:55:36,480 --> 01:55:41,840
students um students you're all amazing

3202
01:55:39,920 --> 01:55:43,840
i hope you knew that already but in case

3203
01:55:41,840 --> 01:55:45,199
you didn't you're amazing you've done

3204
01:55:43,840 --> 01:55:48,000
amazing stuff

3205
01:55:45,199 --> 01:55:50,560
um students who are watching this

3206
01:55:48,000 --> 01:55:52,800
recording and thinking i could never be

3207
01:55:50,560 --> 01:55:55,599
like that

3208
01:55:52,800 --> 01:55:58,000
what what do we do what what what advice

3209
01:55:55,599 --> 01:56:00,480
should we give them amanda

3210
01:55:58,000 --> 01:56:02,000
i think i think that um everybody has a

3211
01:56:00,480 --> 01:56:05,040
different take on things and that you

3212
01:56:02,000 --> 01:56:06,960
can bring something uh any student uh

3213
01:56:05,040 --> 01:56:09,440
can bring something new to this

3214
01:56:06,960 --> 01:56:11,599
conversation about uh the journey that

3215
01:56:09,440 --> 01:56:14,159
they took and the pitfalls that they had

3216
01:56:11,599 --> 01:56:15,040
and and that's encouraging to people and

3217
01:56:14,159 --> 01:56:18,000
also

3218
01:56:15,040 --> 01:56:20,159
um when you do get a little bit um you

3219
01:56:18,000 --> 01:56:23,040
know obsessed about a project in a good

3220
01:56:20,159 --> 01:56:25,520
way then you learn in a way that you

3221
01:56:23,040 --> 01:56:28,400
just don't learn otherwise i think that

3222
01:56:25,520 --> 01:56:30,880
kind of immersion is so important for

3223
01:56:28,400 --> 01:56:32,239
these projects absolutely and i should

3224
01:56:30,880 --> 01:56:36,000
say

3225
01:56:32,239 --> 01:56:38,880
we love hearing about things that didn't

3226
01:56:36,000 --> 01:56:40,800
work just as much as as we love hearing

3227
01:56:38,880 --> 01:56:42,960
about things that did work

3228
01:56:40,800 --> 01:56:45,920
because i think that's something that

3229
01:56:42,960 --> 01:56:48,560
everyone who has tried to make something

3230
01:56:45,920 --> 01:56:51,520
code or otherwise can relate to um and

3231
01:56:48,560 --> 01:56:53,599
it turns out that sharing information

3232
01:56:51,520 --> 01:56:56,480
um of what not to do

3233
01:56:53,599 --> 01:56:57,280
is really hard and really valuable

3234
01:56:56,480 --> 01:56:59,280
yeah

3235
01:56:57,280 --> 01:57:01,040
yeah yeah so if you've got a project

3236
01:56:59,280 --> 01:57:03,280
that you've been working on

3237
01:57:01,040 --> 01:57:05,280
student in the audience listening now

3238
01:57:03,280 --> 01:57:07,440
and everything went wrong and you want

3239
01:57:05,280 --> 01:57:09,280
to tell us the tale of woes

3240
01:57:07,440 --> 01:57:10,480
then you've got two options one is

3241
01:57:09,280 --> 01:57:12,080
submit it to next year's student

3242
01:57:10,480 --> 01:57:14,800
showcase or you can submit it directly

3243
01:57:12,080 --> 01:57:16,480
to the dev oops track which is another

3244
01:57:14,800 --> 01:57:19,840
track that's going on

3245
01:57:16,480 --> 01:57:20,800
um uh but but no stay with us we're cool

3246
01:57:19,840 --> 01:57:22,159
um

3247
01:57:20,800 --> 01:57:24,320
i i just

3248
01:57:22,159 --> 01:57:25,760
amazing these kids are amazing you the

3249
01:57:24,320 --> 01:57:26,639
kids are all right

3250
01:57:25,760 --> 01:57:28,639
yes

3251
01:57:26,639 --> 01:57:30,080
they are all right

3252
01:57:28,639 --> 01:57:31,280
i think we're all right i think i'm

3253
01:57:30,080 --> 01:57:33,280
mostly all right

3254
01:57:31,280 --> 01:57:35,760
thank you to all the teachers who are

3255
01:57:33,280 --> 01:57:37,840
fostering these amazing minds uh out

3256
01:57:35,760 --> 01:57:41,280
there in the world we appreciate you

3257
01:57:37,840 --> 01:57:43,119
there was a lot of love um for specific

3258
01:57:41,280 --> 01:57:44,719
teachers for teachers in general for

3259
01:57:43,119 --> 01:57:46,880
schooling a lot of people wanting a

3260
01:57:44,719 --> 01:57:49,199
refund and wanting to go back to school

3261
01:57:46,880 --> 01:57:52,000
now please i'd like to do the school

3262
01:57:49,199 --> 01:57:53,599
thing again please um i can't say i

3263
01:57:52,000 --> 01:57:55,679
blame them

3264
01:57:53,599 --> 01:57:57,520
yeah yeah

3265
01:57:55,679 --> 01:58:01,360
um teachers

3266
01:57:57,520 --> 01:58:03,840
you are all fantastic um thank you

3267
01:58:01,360 --> 01:58:05,599
so much to everyone who has joined us

3268
01:58:03,840 --> 01:58:07,199
today um

3269
01:58:05,599 --> 01:58:08,800
any last words of wisdom you have about

3270
01:58:07,199 --> 01:58:10,159
the student showcase and then we'll get

3271
01:58:08,800 --> 01:58:13,199
on to some

3272
01:58:10,159 --> 01:58:14,000
closing stuff

3273
01:58:13,199 --> 01:58:17,040
and

3274
01:58:14,000 --> 01:58:20,560
just the most important part

3275
01:58:17,040 --> 01:58:23,040
the duck dance party to get uh to get

3276
01:58:20,560 --> 01:58:24,400
students in person again next year uh

3277
01:58:23,040 --> 01:58:26,080
like i'm

3278
01:58:24,400 --> 01:58:28,239
well fingers crossed like we're going to

3279
01:58:26,080 --> 01:58:31,599
do a dance party to the

3280
01:58:28,239 --> 01:58:33,920
to the gods of of covert um because we

3281
01:58:31,599 --> 01:58:35,360
really do want we really do enjoy

3282
01:58:33,920 --> 01:58:36,800
bringing these students from different

3283
01:58:35,360 --> 01:58:39,760
parts of australia and from different

3284
01:58:36,800 --> 01:58:42,560
walks of life together uh in a sort of

3285
01:58:39,760 --> 01:58:45,119
mini network so that they can um go out

3286
01:58:42,560 --> 01:58:46,800
and and spread the word in the rest of

3287
01:58:45,119 --> 01:58:48,159
the world of their lives yeah i think

3288
01:58:46,800 --> 01:58:50,480
it's really important to get them

3289
01:58:48,159 --> 01:58:52,560
together so i really like how inclusive

3290
01:58:50,480 --> 01:58:53,760
the chat is and how the students have

3291
01:58:52,560 --> 01:58:55,440
been

3292
01:58:53,760 --> 01:58:57,840
asked questions like experts and we

3293
01:58:55,440 --> 01:58:59,920
really appreciate how supportive this

3294
01:58:57,840 --> 01:59:01,199
community is but i'm really looking

3295
01:58:59,920 --> 01:59:03,280
forward to some

3296
01:59:01,199 --> 01:59:05,040
uh face-to-face action

3297
01:59:03,280 --> 01:59:06,159
indeed indeed

3298
01:59:05,040 --> 01:59:07,360
um

3299
01:59:06,159 --> 01:59:09,360
all right

3300
01:59:07,360 --> 01:59:10,719
but don't take that as a

3301
01:59:09,360 --> 01:59:12,719
oh well i'm not sure if i could get

3302
01:59:10,719 --> 01:59:14,480
there so i won't go ahead and try and

3303
01:59:12,719 --> 01:59:15,440
build that cool thing because i don't

3304
01:59:14,480 --> 01:59:17,520
even know

3305
01:59:15,440 --> 01:59:20,480
just go and build the cool thing

3306
01:59:17,520 --> 01:59:23,199
build it anyway and tell us about it

3307
01:59:20,480 --> 01:59:25,920
yeah logistics logistics are next year's

3308
01:59:23,199 --> 01:59:28,800
problem and i please everyone go and

3309
01:59:25,920 --> 01:59:31,040
make that a really horrible challenging

3310
01:59:28,800 --> 01:59:32,159
problem for us um that would be

3311
01:59:31,040 --> 01:59:32,960
amazing

3312
01:59:32,159 --> 01:59:34,000
yeah

3313
01:59:32,960 --> 01:59:35,360
you know next year we're just going to

3314
01:59:34,000 --> 01:59:36,400
have to cancel all of our talks and the

3315
01:59:35,360 --> 01:59:38,719
whole thing is going to be student

3316
01:59:36,400 --> 01:59:40,000
showcase so it's like

3317
01:59:38,719 --> 01:59:42,150
we can't do that i learned too much

3318
01:59:40,000 --> 01:59:45,870
today i have to sit down

3319
01:59:42,150 --> 01:59:45,870
[Laughter]