-
Notifications
You must be signed in to change notification settings - Fork 22
/
chapter3.html
583 lines (493 loc) · 32.6 KB
/
chapter3.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
<!DOCTYPE html>
<html lang="en" dir="ltr">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0, shrink-to-fit=no">
<!-- <meta http-equiv="X-UA-Compatible" content="IE=edge"> -->
<!-- Social media -->
<meta name="twitter:card" content="summary_large_image" />
<meta name="twitter:site" content="@OKAI" />
<!-- BB Change this--><meta name="twitter:title" content="Chapter 3: The Building Block of Deep Learning" />
<!-- CC Change this--><meta name="twitter:description" content="In this chapter, we take a look at a perceptron, which composes neural networks." />
<!-- AA Change this--><meta name="twitter:image" content="https://okai.brown.edu/img/share/en/en_3.png" />
<!-- Change this--><meta property="og:url" content="https://okai.brown.edu/chapter3.html" />
<meta property="og:type" content="article" />
<!-- BB Change this--><meta property="og:title" content="Chapter 3: The Building Block of Deep Learning" />
<!-- CC Change this--><meta property="og:description" content="In this chapter, we take a look at a perceptron, which composes neural networks." />
<!-- AA Change this--><meta property="og:image" content="https://okai.brown.edu/img/share/en/en_3.png" />
<!-- ................... -->
<!-- Google Fonts -->
<link href="https://fonts.googleapis.com/css?family=Nunito:400,700" rel="stylesheet">
<!-- Font Awesome -->
<link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.7.0/css/all.css" integrity="sha384-lZN37f5QGtY3VHgisS14W3ExzMWZxybE1SJSEsQp9S+oqd12jhcu+A56Ebc1zFSJ"
crossorigin="anonymous">
<!-- Bootstrap CSS CDN -->
<link rel="stylesheet" href="https://stackpath.bootstrapcdn.com/bootstrap/4.1.0/css/bootstrap.min.css"
integrity="sha384-9gVQ4dYFwwWSjIDZnLEWnxCjeSWFphJiwGPXr1jddIhOegiu1FwO5qRGvFXOdJZ4"
crossorigin="anonymous">
<!-- Scrollbar Custom CSS -->
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/malihu-custom-scrollbar-plugin/3.1.5/jquery.mCustomScrollbar.min.css">
<!-- My stylesheet -->
<link rel="stylesheet" href="./css/general.css">
<link rel="stylesheet" href="./css/chapter3.css">
<!-- language annotations -->
<link rel="alternate" hreflang="zh-Hans" href="https://okai.brown.edu/zh/chapter3.html" />
<link rel="alternate" hreflang="en" href="https://okai.brown.edu/chapter3.html" />
<link rel="canonical" href="https://okai.brown.edu/chapter3.html" />
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=UA-133914635-1"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag() {
dataLayer.push(arguments);
}
gtag('js', new Date());
gtag('config', 'UA-133914635-1');
</script>
<title>Chapter 3 - OKAI</title>
</head>
<body>
<!-- scrollbar -->
<div class="scrollBar" id="myBar"></div>
<!-- navbar place holder -->
<div id="nav-placeholder"></div>
<!-- Chapter 3 content beginning -->
<!-- landing section -->
<div id="openingText" class="d-flex align-items-center justify-content-center">
<!-- start of row 1 -->
<div class="row align-items-center justify-content-center">
<!-- col 1 -->
<div class="d-none d-md-block col-1 arrowIcon" style="position: absolute; left: 5px;">
<a href="./chapter2.html" data-toggle="tooltip" title="Chapter 2"><i class="fas fa-caret-left fa-2x"></i></a>
</div>
<!-- col 2 -->
<div class="col-10 col-md-4 order-2 order-md-1 text-center">
<div class="text-left">
<p>Chapter 3</p>
<h2>The Building Block of Deep Learning</h2>
<p>Complex networks, like the ones used in deep learning, are built from many
individual components. In this chapter, we take a look at a perceptron,
which composes neural networks. Keep scrolling!
</p>
</div>
</div>
<!-- col 3 -->
<div class="col-md-5 order-1 order-md-2">
<div id="openingLottie" class="openingAnimation mx-auto"></div>
</div>
<!-- col-4 -->
<div class="d-none d-md-block col-1 arrowIcon" style="position: absolute; right: 5px;">
<a href="./chapter4.html" data-toggle="tooltip" title="Chapter 4"><i class="fas fa-caret-right fa-2x"></i></a>
</div>
</div>
<!-- end of row 1 -->
<!-- start of row 2 -->
<div class="" style="position: absolute; bottom: 20px;">
<div>
<p class="iconHeight" id="scrollIcon">Scroll</p>
<i class="iconHeight fas fa-angle-down fa-lg"></i>
</div>
<!-- end of row 2 -->
</div>
</div>
<!-- end of landing section -->
<div id="neuronLottie" class="animation animLeft upperLayer"></div>
<div id="activationLottie" class="animation animLeft threeLayer"></div>
<!-- text begins -->
<div id="brainText" class="bg-1grey scene">
<div class="text right">
<p>Early neural network researchers in the 1950s took inspirations from a neuron, the
building block of brains, to create an algorithm called the perceptron.
</p>
<p>To understand what a perceptron is, let’s first take a look at a neuron.
</p>
</div>
</div>
<div id="nerveText" class="bg-1grey scene">
<div class="text right">
<p>A neuron is composed of 3 main parts:</p>
<p>Dendrites: Receivers of neural information that takes in stimuli from other neurons.</p>
<p>Soma: The “core” of a neuron where the nucleus is located; it decides if a signal
will be sent to other neurons.</p>
<p>Axon: Long extending neck of the neuron that passes any signal to any of the
neighbor neurons.</p>
</div>
</div>
<div id="neuron1Text" class="scene bg-2grey upperLayer">
<div class="text left">
<p>The perceptron, similarly, also has 3 major parts:</p>
<p>Inputs: inputs take in signals in the form of numbers and pass them to the
perceptron.</p>
<p>Perceptron: the perceptron takes in the inputs and conducts some calculations on
them.</p>
<p>
Output: the output of the perceptron spits out the result of the calculations.
</p>
</li>
</ul>
</div>
</div>
<div id="neuron2Text" class="scene bg-2grey upperLayer">
<div class="text left">
<p>Between inputs and perceptrons, the axon-to-dendrite connection is simplified to an
edge, and signals transmitted are represented with numbers.<br>
The input on top here is connected to the perceptron with a thicker edge, thus the
signal is amplified, in this case by a factor of 2, when it reaches the neuron.
</p>
</div>
</div>
<div id="neuron3Text" class="scene bg-2grey upperLayer">
<div class="text left">
<p>However, the input at the bottom has a thinner connection to the neuron, in this
case reducing the signal to two thirds of what it was. The perceptron then sums the
received signals: 2 from the top and 2 from the bottom, resulting in a final signal
of 4.
<br>
The thickness of the edge is called <strong class="highlight" data-toggle="tooltip"
title="A numeric value associated with an edge. It can amplify, reduce, or invert signals.">weight</strong>.
</p>
</div>
</div>
<div id="neuron4Text" class="scene bg-2grey upperLayer">
<div class="text left">
<p>Next, the perceptron processes the signal by "shaking" itself. In this case, doing so
increased the signal by 1.
<br>
The “shakiness” of a perceptron is called <strong class="highlight" data-toggle="tooltip"
title="A numeric value added onto the weighted sum of all input signals.">bias</strong>,
which can either increase or decrease the value of the signal by any set amount.
</p>
</div>
</div>
<div id="neuron5Text" class="scene bg-2grey upperLayer">
<div class="text left">
<p>Finally, an <strong class="highlight" data-toggle="tooltip" title="A function that is used to control the final output of a perceptron.">
activation function</strong> is used to decide how strong a signal the
perceptron should transmit to its neighbors. More on this issue soon!
</p>
<p>
The sigmoid function, a type of activation functions, is used here. It maps the
signal value into the range of 0-1. In our case, 5 is being mapped to 0.99, which
becomes the final output.
</p>
</div>
</div>
<!-- activation function animation text section -->
<div id="appanana1Text" class="scene topLayer bg-3grey">
<div class="text right">
<h3>
Activation Functions
</h3>
<hr>
<p>
Early researchers tried to use these perceptrons to determine how likely an image
contains a human face, but they quickly ran into a problem: the
resulting probability of the model was often above 100% or below 0%.
</p>
</div>
</div>
<div id="appanana2Text" class="scene topLayer bg-3grey">
<div class="text right">
<p>
To tackle this problem, the researchers put an activation function behind the
output of the perceptron, which could control the output range of that perceptron.
The sigmoid function and ReLU are examples of activation functions that are
commonly used to achieve different output ranges.
</p>
</div>
</div>
<div id="appanana3Text" class="scene topLayer bg-3grey">
<div class="text right">
<p>
Sigmoid calculates the probability that something is true. For example, if you
want to determine if an image is a cat or not, you would use a sigmoid activation
function, which transforms the output value into a probability (a number between 0
and 1).
</p>
</div>
</div>
<div id="appanana4Text" class="scene topLayer bg-3grey">
<div class="text right">
<p>
ReLU makes sure that the final output is non-negative, and is commonly used in
multilayer networks. Unlike sigmoid whose main purpose is to limit output range,
ReLU is used to break linearity so that the multilayer networks can learn more
complex concepts. We will talk about this notion in depth in Chapter 5.
</p>
</div>
</div>
<!-- ending section -->
<div id="endingText" class="lastScene row align-items-center justify-content-around fourLayer">
<!-- col 1 -->
<div class="d-none d-md-block col-1 arrowIcon" style="position: absolute; left: 5px;">
<a href="./chapter2.html" data-toggle="tooltip" title="Chapter 2"><i class="fas fa-caret-left fa-2x"></i></a>
</div>
<div class="col-12 col-md-10">
<div class="row justify-content-around">
<div class="col-10 col-md-5 fourLayer align-self-start">
<div class="text-left align-self-start">
<h3><strong>Summary</strong></h3>
<p>Up to this point, we have looked at how neurons inspired the perceptron
algorithm, and what components make up the perceptron. In the next
chapter, we will discuss how data can be used to teach perceptrons and
other more complex neural networks.
</p>
<h3><strong>Further Reading</strong></h3>
<p>
<a href="https://towardsdatascience.com/what-the-hell-is-perceptron-626217814f53"
target="_blank">
What the Hell is Perceptron?</a>
<br><a href="https://www.cs.cmu.edu/afs/cs.cmu.edu/academic/class/15381-f01/www/handouts/110601.pdf"
target="_blank">
Perceptrons and Neural Networks</a>
<!-- <br><a href="https://cs.stanford.edu/people/eroberts/courses/soco/projects/neural-networks/Architecture/feedforward.html" target="_blank">
Neural Networks - Architecture - Stanford CS</a> -->
</p>
</div>
</div>
<div class="col-10 col-md-5 fourLayer align-self-start">
<div class="text-left">
<h3><strong>Glossary</strong></h3>
<p>
<strong>Perceptron</strong>: A neuron-inspired algorithm that is the
building block of more complex neural networks.
<br><br>
<strong>Weight</strong>: A numeric value associated with an edge (the
connection between two perceptrons). It can amplify, reduce, or invert
signals.
<br><br>
<strong>Bias</strong>: A numeric value added onto the weighted sum of
all input signals.
<br><br>
<strong>Activation Function</strong>: A function that is used to
control the final output of a perceptron.
</p>
</div>
</div>
<!-- mobile nav button groups -->
<div class="col-10 d-md-none fourLayer justify-content-start my-3">
<div class="btn-group d-flex" id="nextButtonGroup" role="group">
<a class="btn btn-warning w-50" href="./chapter2.html"
role="button" id="prevButton">Prev Chapter</a>
<span class="btn-warning btn-separator"></span>
<a class="btn btn-warning w-50" href="./chapter4.html"
role="button" id="nextButton">Next Chapter</a>
</div>
</div>
<!-- end of inner row -->
</div>
<!-- end of outer col-10 -->
</div>
<!-- col-4 -->
<div class="d-none d-md-block col-1 arrowIcon" style="position: absolute; right: 5px;">
<a href="./chapter4.html" data-toggle="tooltip" title="Chapter 4"><i class="fas fa-caret-right fa-2x"></i></a>
</div>
<!-- end of ending section -->
</div>
<!-- end of text area -->
<div class="overlay"></div>
<!-- beginning of perceptronAnimation -->
<div id="perceptronAnimation" class="animation animRight upperLayer">
<svg id="neuronSys" class="neuronSVG" version="1.1" xmlns="https://www.w3.org/2000/svg"
viewBox="0 0 600 600">
<defs>
<filter color-interpolation-filters="sRGB" id="goo">
<feGaussianBlur in="SourceGraphic" stdDeviation="8" result="blur" />
<feColorMatrix in="blur" mode="matrix" values="1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 21 -9"
result="cm" />
<feBlend />
</filter>
</defs>
<rect id="perceptronBG" fill="#f1f1f1" x="0" y="0" width="600" height="600" />
<g id="neuronGroup" filter="url(#goo)">
<line id="dendrites1" class="dendrites" x1="186" y1="214" x2="266" y2="275" />
<line id="dendrites2" class="dendrites" x1="184" y1="387" x2="268" y2="324" />
<line class="axon" x1="407" y1="300" x2="340" y2="300" />
<g id="neuron">
<circle class="whitespace" cx="300" cy="300" r="41" />
</g>
<g class="input" id="input1">
<circle class="whitespace" cx="162" cy="196" r="29" />
</g>
<g class="input" id="input2">
<circle class="whitespace" cx="162" cy="404" r="29" />
</g>
<g class="output" id="output">
<circle class="whitespace" cx="434" cy="301" r="29.5" />
</g>
<line id="inTransit1" class="inTransit" x1="186" y1="214" x2="266" y2="275" />
<line id="inTransit2" class="inTransit" x1="184" y1="387" x2="268" y2="324" />
<g class="signal" id="topSignal">
<circle class="st1" cx="162" cy="195" r="20" />
</g>
<g class="signal" id="botSignal">
<circle class="st1" cx="162" cy="403.7" r="20" />
</g>
<circle id="cenSig" cx="300" cy="300" r="30" />
<circle id="cenSig3" cx="300" cy="300" r="30" />
<circle id="outSig" cx="434" cy="301" r="25" />
</g>
<g id="num1" class="num">
<path d="M163.7,185.7c0.4,0.4,0.5,0.8,0.5,1.3V205c0,0.5-0.2,1-0.6,1.3c-0.4,0.4-0.9,0.5-1.5,0.5c-0.6,0-1.1-0.2-1.4-0.5
c-0.4-0.4-0.6-0.8-0.6-1.3v-14.7l-1.9,1.2c-0.3,0.2-0.6,0.3-1,0.3c-0.5,0-1-0.2-1.3-0.6s-0.5-0.8-0.5-1.3c0-0.3,0.1-0.6,0.3-0.9
c0.2-0.3,0.4-0.5,0.7-0.7l4.7-2.8c0.4-0.2,0.8-0.3,1.3-0.3C162.9,185.1,163.3,185.3,163.7,185.7z" />
</g>
<g id="num3" class="num">
<path d="M166.2,402.2c0.9,0.6,1.6,1.4,2,2.4c0.5,1,0.7,2.1,0.7,3.2c0,1.5-0.3,2.8-1,4c-0.7,1.1-1.6,2-2.9,2.6
c-1.2,0.6-2.7,0.9-4.3,0.9c-0.7,0-1.5-0.1-2.2-0.3c-0.7-0.2-1.4-0.4-2-0.7c-0.8-0.4-1.1-1-1.1-1.7c0-0.5,0.2-1,0.5-1.4
c0.4-0.4,0.8-0.6,1.3-0.6c0.4,0,0.9,0.2,1.3,0.5c0.9,0.7,1.9,1,2.9,1c0.6,0,1.3-0.2,1.8-0.5c0.6-0.3,1-0.8,1.4-1.4
c0.4-0.6,0.5-1.3,0.5-2.1c0-1.1-0.3-2-1-2.6c-0.7-0.6-1.5-0.9-2.4-0.9c-0.4,0-0.8,0-1.1,0.1c-0.3,0.1-0.5,0.1-0.6,0.2
c-0.5,0.2-1,0.2-1.3,0.2c-0.5,0-0.8-0.2-1.1-0.5c-0.3-0.3-0.4-0.7-0.4-1.2c0-0.3,0.1-0.5,0.2-0.8c0.1-0.2,0.3-0.5,0.5-0.7l4.8-5.2
h-6.1c-0.5,0-0.9-0.2-1.2-0.5c-0.3-0.3-0.5-0.7-0.5-1.2c0-0.5,0.2-0.9,0.5-1.2c0.3-0.3,0.7-0.5,1.2-0.5h9.5c0.6,0,1.1,0.2,1.4,0.5
c0.3,0.3,0.5,0.8,0.5,1.3c0,0.5-0.2,1-0.7,1.5l-4.2,4.6C164.3,401.2,165.3,401.5,166.2,402.2z" />
</g>
<g class="num">
<path id="num2" d="M309.6,311.7c0.5,0.5,0.7,1.1,0.7,1.8c0,0.7-0.2,1.3-0.7,1.7c-0.5,0.5-1.1,0.7-1.8,0.7h-15.3c-0.8,0-1.4-0.2-1.8-0.7
c-0.5-0.5-0.7-1.1-0.7-1.8c0-0.7,0.3-1.4,0.8-2l9.8-10.5c1.1-1.2,2-2.4,2.6-3.6c0.6-1.2,1-2.3,1-3.4c0-1.4-0.5-2.5-1.4-3.5
c-1-1-2.1-1.5-3.3-1.5c-0.9,0-1.7,0.3-2.6,0.9s-1.7,1.4-2.4,2.3c-0.5,0.7-1.2,1.1-2.1,1.1c-0.7,0-1.3-0.3-1.8-0.8
c-0.6-0.5-0.8-1.1-0.8-1.7c0-0.5,0.1-0.9,0.5-1.4c0.3-0.5,0.7-1,1.3-1.6c1.1-1.1,2.5-2.1,4-2.7s2.9-1,4.3-1c2,0,3.7,0.4,5.2,1.3
c1.5,0.8,2.7,2,3.5,3.5c0.8,1.5,1.2,3.1,1.2,5c0,1.8-0.5,3.7-1.5,5.6c-1,1.9-2.3,3.8-3.9,5.6l-5.7,6.1h9.3
C308.5,311,309.1,311.3,309.6,311.7z" />
</g>
<g class="num">
<path id="num4" d="M311.4,303c0.5,0.5,0.7,1.1,0.7,1.8c0,0.7-0.2,1.3-0.7,1.7c-0.5,0.5-1.1,0.7-1.8,0.7h-1.9v5.8c0,0.8-0.3,1.4-0.8,2
c-0.5,0.5-1.2,0.8-2,0.8c-0.8,0-1.4-0.3-1.9-0.8c-0.5-0.5-0.8-1.2-0.8-2v-5.8h-11.7c-0.7,0-1.3-0.3-1.8-0.8
c-0.6-0.5-0.9-1.1-0.9-1.9c0-0.7,0.2-1.3,0.7-1.8l14.2-17.4c0.6-0.7,1.3-1.1,2.3-1.1c0.7,0,1.4,0.3,1.9,0.8c0.5,0.5,0.8,1.2,0.8,2
v15.3h1.9C310.3,302.3,310.9,302.5,311.4,303z M295,302.3h7.3v-9L295,302.3z" />
</g>
<g class="num">
<path id="numBig3" d="M306.2,297c1.3,0.9,2.3,2.1,2.9,3.5c0.7,1.5,1,3,1,4.7c0,2.2-0.5,4.1-1.5,5.7c-1,1.6-2.4,2.9-4.2,3.8s-3.9,1.3-6.3,1.3
c-1.1,0-2.2-0.1-3.2-0.4s-2-0.6-2.8-1.1c-1.1-0.6-1.7-1.4-1.7-2.4c0-0.7,0.3-1.4,0.8-2c0.5-0.6,1.1-0.9,1.8-0.9
c0.6,0,1.2,0.2,1.9,0.7c1.3,1,2.7,1.5,4.3,1.5c0.9,0,1.8-0.2,2.7-0.7c0.8-0.5,1.5-1.1,2-2c0.5-0.9,0.8-1.9,0.8-3
c0-1.6-0.5-2.9-1.5-3.8c-1-0.9-2.2-1.4-3.5-1.4c-0.6,0-1.1,0.1-1.6,0.2c-0.5,0.1-0.7,0.2-0.8,0.2c-0.8,0.2-1.4,0.4-1.9,0.4
c-0.7,0-1.2-0.2-1.6-0.7c-0.4-0.5-0.6-1-0.6-1.7c0-0.4,0.1-0.8,0.2-1.1c0.1-0.3,0.4-0.7,0.8-1.1l7-7.6h-8.9c-0.7,0-1.3-0.2-1.8-0.7
c-0.5-0.5-0.7-1.1-0.7-1.8c0-0.7,0.2-1.3,0.7-1.7c0.5-0.5,1.1-0.7,1.8-0.7h13.8c0.9,0,1.6,0.2,2.1,0.7c0.5,0.5,0.7,1.1,0.7,2
c0,0.7-0.4,1.4-1.1,2.2l-6,6.7C303.4,295.6,304.9,296.1,306.2,297z" />
<g class="num">
<path id="numBig5" d="M305,295.7c1.6,0.9,2.9,2.2,3.8,3.8c0.9,1.6,1.4,3.4,1.4,5.3c0,2-0.6,3.9-1.7,5.6c-1.1,1.7-2.7,3.1-4.5,4.1
c-1.9,1-3.9,1.5-6.1,1.5c-1.2,0-2.4-0.2-3.6-0.5c-1.2-0.4-2.3-0.9-3-1.5c-0.4-0.3-0.7-0.7-0.9-1.3s-0.4-1-0.4-1.5
c0-0.5,0.2-1,0.7-1.5c0.5-0.4,1.1-0.7,1.8-0.7c0.5,0,1.2,0.3,2,0.9c1.3,0.8,2.4,1.2,3.3,1.2c1.2,0,2.4-0.3,3.5-0.9
c1.1-0.6,1.9-1.3,2.6-2.3c0.6-1,1-2,1-3.1c0-1.6-0.6-2.9-1.7-4c-1.1-1-2.4-1.5-4-1.5c-0.7,0-1.3,0.1-1.8,0.3
c-0.5,0.2-1.2,0.5-1.9,0.8c-0.5,0.3-1,0.5-1.3,0.6c-0.3,0.1-0.6,0.2-1,0.2c-1.2,0-2-0.3-2.5-0.9c-0.5-0.6-0.7-1.4-0.7-2.3
c0-0.3,0-0.5,0-0.6l1.4-11.2c0.1-0.6,0.4-1.1,0.9-1.6c0.5-0.4,1.1-0.6,1.8-0.6h13c0.7,0,1.3,0.2,1.8,0.7s0.7,1.1,0.7,1.8
c0,0.7-0.2,1.3-0.7,1.7c-0.5,0.5-1.1,0.7-1.8,0.7h-11.2l-0.9,6.5c0.6-0.3,1.3-0.6,2.2-0.8c0.9-0.2,1.7-0.3,2.5-0.3
C301.6,294.4,303.4,294.8,305,295.7z" />
</g>
<g class="num">
<g id="numOut5">
<path d="M418.8,309.5c-0.26-0.26-0.39-0.63-0.39-1.11v-0.43c0-0.48,0.13-0.85,0.39-1.11c0.26-0.26,0.63-0.39,1.11-0.39h0.33
c0.48,0,0.85,0.13,1.11,0.39c0.26,0.26,0.39,0.63,0.39,1.11v0.43c0,0.48-0.13,0.85-0.39,1.11c-0.26,0.26-0.63,0.39-1.11,0.39h-0.33
C419.43,309.89,419.06,309.76,418.8,309.5z" />
<path d="M433.09,293.05c0.9,0.64,1.59,1.48,2.07,2.52s0.73,2.15,0.73,3.34c0,1.22-0.21,2.55-0.64,4.01s-1.2,2.84-2.33,4.14
c-1.13,1.3-2.65,2.23-4.59,2.77c-0.2,0.05-0.39,0.08-0.58,0.08c-0.43,0-0.79-0.11-1.08-0.34s-0.42-0.58-0.42-1.06
c0-0.37,0.12-0.69,0.36-0.96c0.24-0.28,0.56-0.47,0.96-0.59c1.05-0.3,2.02-0.83,2.91-1.6s1.55-1.68,1.96-2.75
c-0.83,0.72-1.78,1.07-2.85,1.07c-0.87,0-1.72-0.26-2.56-0.79c-0.84-0.52-1.53-1.23-2.06-2.13c-0.53-0.89-0.8-1.85-0.8-2.89
c0-1.05,0.26-2.02,0.79-2.91c0.53-0.89,1.23-1.6,2.11-2.11c0.88-0.52,1.86-0.78,2.92-0.78
C431.16,292.09,432.19,292.41,433.09,293.05z M432.02,300.06c0.54-0.58,0.81-1.31,0.81-2.17s-0.27-1.59-0.81-2.18
c-0.54-0.58-1.21-0.88-2.01-0.88c-0.78,0-1.45,0.29-2,0.88c-0.55,0.58-0.83,1.31-0.83,2.18s0.28,1.59,0.83,2.17
c0.55,0.58,1.22,0.88,2,0.88C430.81,300.94,431.48,300.65,432.02,300.06z" />
<path d="M446.79,293.05c0.9,0.64,1.59,1.48,2.07,2.52s0.73,2.15,0.73,3.34c0,1.22-0.21,2.55-0.64,4.01s-1.2,2.84-2.33,4.14
c-1.13,1.3-2.65,2.23-4.59,2.77c-0.2,0.05-0.39,0.08-0.58,0.08c-0.43,0-0.79-0.11-1.08-0.34s-0.42-0.58-0.42-1.06
c0-0.37,0.12-0.69,0.36-0.96c0.24-0.28,0.56-0.47,0.96-0.59c1.05-0.3,2.02-0.83,2.91-1.6s1.55-1.68,1.96-2.75
c-0.83,0.72-1.78,1.07-2.85,1.07c-0.87,0-1.72-0.26-2.56-0.79c-0.84-0.52-1.53-1.23-2.06-2.13c-0.53-0.89-0.8-1.85-0.8-2.89
c0-1.05,0.26-2.02,0.79-2.91c0.53-0.89,1.23-1.6,2.11-2.11c0.88-0.52,1.86-0.78,2.92-0.78
C444.86,292.09,445.89,292.41,446.79,293.05z M445.73,300.06c0.54-0.58,0.81-1.31,0.81-2.17s-0.27-1.59-0.81-2.18
c-0.54-0.58-1.21-0.88-2.01-0.88c-0.78,0-1.45,0.29-2,0.88c-0.55,0.58-0.83,1.31-0.83,2.18s0.28,1.59,0.83,2.17
c0.55,0.58,1.22,0.88,2,0.88C444.51,300.94,445.18,300.65,445.73,300.06z" />
</g>
</g>
<g id="ReLU">
<path class="st0" d="M354.65,302.81c-0.5-0.15-0.91-0.35-1.23-0.61c-0.17-0.13-0.25-0.3-0.25-0.53c0-0.15,0.04-0.29,0.14-0.4
s0.2-0.17,0.33-0.17c0.13,0,0.28,0.05,0.45,0.15c0.32,0.23,0.65,0.39,0.99,0.5s0.73,0.15,1.14,0.15c0.54,0,0.95-0.09,1.23-0.28
s0.43-0.46,0.43-0.82c0-0.28-0.13-0.49-0.4-0.64s-0.71-0.3-1.34-0.43c-0.65-0.14-1.17-0.3-1.57-0.48c-0.4-0.18-0.71-0.42-0.91-0.7
s-0.31-0.65-0.31-1.09c0-0.46,0.13-0.86,0.38-1.22s0.61-0.64,1.07-0.84s0.97-0.3,1.55-0.3c1.05,0,1.9,0.28,2.56,0.84
c0.1,0.08,0.16,0.16,0.2,0.24c0.04,0.08,0.06,0.17,0.06,0.29c0,0.15-0.05,0.29-0.14,0.4s-0.2,0.17-0.32,0.17
c-0.07,0-0.14-0.01-0.2-0.03s-0.14-0.06-0.25-0.12c-0.3-0.22-0.59-0.38-0.88-0.49c-0.28-0.11-0.63-0.16-1.04-0.16
c-0.5,0-0.89,0.1-1.18,0.3s-0.43,0.48-0.43,0.84c0,0.3,0.13,0.53,0.38,0.69c0.26,0.16,0.69,0.31,1.31,0.45
c0.65,0.15,1.18,0.31,1.59,0.49c0.41,0.18,0.72,0.41,0.94,0.69c0.22,0.28,0.32,0.63,0.32,1.04c0,0.45-0.13,0.86-0.38,1.2
c-0.25,0.35-0.61,0.62-1.07,0.81c-0.46,0.19-0.98,0.29-1.58,0.29C355.69,303.03,355.15,302.96,354.65,302.81z" />
<path class="st0" d="M360.55,296.26c-0.15-0.14-0.22-0.31-0.22-0.53c0-0.22,0.07-0.4,0.22-0.53c0.15-0.13,0.34-0.2,0.59-0.2
c0.24,0,0.44,0.07,0.59,0.2c0.15,0.13,0.23,0.31,0.23,0.53c0,0.22-0.07,0.4-0.22,0.53c-0.15,0.14-0.34,0.2-0.59,0.2
C360.9,296.46,360.7,296.4,360.55,296.26z M360.66,302.83c-0.13-0.11-0.19-0.28-0.19-0.49v-4.25c0-0.21,0.06-0.37,0.19-0.48
c0.13-0.11,0.29-0.16,0.49-0.16c0.2,0,0.36,0.05,0.49,0.16c0.13,0.11,0.2,0.27,0.2,0.48v4.25c0,0.21-0.07,0.38-0.2,0.49
c-0.13,0.11-0.3,0.17-0.49,0.17C360.95,303,360.79,302.94,360.66,302.83z" />
<path class="st0" d="M368.31,297.62c0.13,0.12,0.19,0.27,0.19,0.47v4.34c0,0.84-0.24,1.48-0.7,1.91c-0.47,0.43-1.16,0.65-2.08,0.65
c-0.8,0-1.47-0.15-2-0.45c-0.27-0.16-0.41-0.36-0.41-0.58c0-0.16,0.04-0.29,0.12-0.39c0.08-0.1,0.18-0.14,0.3-0.14
c0.07,0,0.15,0.02,0.24,0.05c0.1,0.03,0.19,0.07,0.3,0.12c0.26,0.11,0.5,0.19,0.71,0.25s0.45,0.09,0.72,0.09
c0.96,0,1.44-0.47,1.44-1.41v-0.7c-0.16,0.3-0.4,0.53-0.7,0.7c-0.31,0.17-0.66,0.25-1.06,0.25c-0.47,0-0.89-0.11-1.25-0.34
c-0.36-0.22-0.65-0.54-0.85-0.94c-0.21-0.4-0.31-0.87-0.31-1.4c0-0.53,0.1-1,0.3-1.4c0.2-0.41,0.49-0.72,0.85-0.95
c0.37-0.22,0.78-0.34,1.25-0.34c0.4,0,0.75,0.08,1.06,0.25c0.31,0.17,0.54,0.4,0.7,0.7v-0.29c0-0.19,0.06-0.34,0.19-0.46
c0.12-0.12,0.29-0.18,0.49-0.18C368.02,297.44,368.18,297.5,368.31,297.62z M366.76,301.28c0.25-0.29,0.37-0.68,0.37-1.18
c0-0.51-0.12-0.9-0.37-1.19c-0.25-0.29-0.59-0.43-1.03-0.43c-0.43,0-0.77,0.14-1.02,0.43c-0.25,0.29-0.37,0.68-0.37,1.19
c0,0.5,0.13,0.89,0.37,1.18s0.59,0.43,1.02,0.43C366.17,301.71,366.52,301.57,366.76,301.28z" />
<path class="st0" d="M377.65,297.94c0.31,0.35,0.47,0.91,0.47,1.69v2.71c0,0.21-0.06,0.38-0.19,0.49s-0.29,0.17-0.5,0.17
c-0.2,0-0.36-0.06-0.49-0.17c-0.13-0.11-0.19-0.28-0.19-0.49v-2.71c0-0.4-0.07-0.69-0.21-0.86c-0.14-0.17-0.36-0.26-0.65-0.26
c-0.36,0-0.65,0.12-0.86,0.37c-0.21,0.25-0.32,0.58-0.32,1.01v2.45c0,0.21-0.06,0.38-0.19,0.49c-0.13,0.11-0.29,0.17-0.49,0.17
c-0.2,0-0.36-0.06-0.5-0.17s-0.2-0.28-0.2-0.49v-2.71c0-0.4-0.07-0.69-0.21-0.86c-0.14-0.17-0.36-0.26-0.65-0.26
c-0.36,0-0.64,0.12-0.85,0.37c-0.21,0.25-0.31,0.58-0.31,1.01v2.45c0,0.21-0.07,0.38-0.2,0.49c-0.13,0.11-0.3,0.17-0.49,0.17
c-0.2,0-0.36-0.06-0.49-0.17c-0.13-0.11-0.19-0.28-0.19-0.49v-4.26c0-0.21,0.07-0.36,0.2-0.47c0.13-0.11,0.3-0.16,0.5-0.16
c0.19,0,0.35,0.05,0.47,0.16c0.12,0.11,0.18,0.26,0.18,0.46v0.26c0.17-0.29,0.39-0.52,0.68-0.67c0.28-0.15,0.61-0.23,0.97-0.23
c0.81,0,1.35,0.34,1.62,1.01c0.16-0.31,0.4-0.55,0.71-0.74c0.32-0.18,0.67-0.28,1.07-0.28
C376.89,297.42,377.34,297.59,377.65,297.94z" />
<path class="st0" d="M380.52,302.68c-0.42-0.23-0.74-0.55-0.96-0.98s-0.34-0.92-0.34-1.48c0-0.56,0.11-1.06,0.34-1.48
c0.23-0.42,0.55-0.75,0.96-0.97c0.41-0.23,0.9-0.34,1.45-0.34s1.03,0.11,1.45,0.34s0.73,0.55,0.96,0.97
c0.22,0.42,0.34,0.91,0.34,1.48c0,0.56-0.11,1.06-0.34,1.48s-0.54,0.75-0.96,0.98c-0.41,0.23-0.9,0.34-1.45,0.34
S380.94,302.91,380.52,302.68z M382.98,301.51c0.24-0.29,0.35-0.73,0.35-1.3c0-0.56-0.12-1-0.35-1.29
c-0.23-0.3-0.57-0.45-1.01-0.45s-0.78,0.15-1.02,0.45c-0.24,0.3-0.36,0.73-0.36,1.29c0,0.57,0.12,1,0.35,1.3
c0.24,0.29,0.57,0.44,1.01,0.44C382.41,301.95,382.75,301.81,382.98,301.51z" />
<path class="st0" d="M385.93,296.26c-0.15-0.14-0.22-0.31-0.22-0.53c0-0.22,0.07-0.4,0.22-0.53c0.15-0.13,0.34-0.2,0.59-0.2
c0.24,0,0.44,0.07,0.59,0.2c0.15,0.13,0.23,0.31,0.23,0.53c0,0.22-0.07,0.4-0.22,0.53c-0.15,0.14-0.34,0.2-0.59,0.2
C386.27,296.46,386.08,296.4,385.93,296.26z M386.03,302.83c-0.13-0.11-0.19-0.28-0.19-0.49v-4.25c0-0.21,0.06-0.37,0.19-0.48
c0.13-0.11,0.29-0.16,0.49-0.16c0.2,0,0.36,0.05,0.49,0.16c0.13,0.11,0.2,0.27,0.2,0.48v4.25c0,0.21-0.07,0.38-0.2,0.49
c-0.13,0.11-0.3,0.17-0.49,0.17C386.33,303,386.16,302.94,386.03,302.83z" />
<path class="st0" d="M393.63,295.3c0.13,0.12,0.2,0.27,0.2,0.46v6.58c0,0.2-0.06,0.36-0.19,0.47c-0.12,0.12-0.29,0.18-0.5,0.18
c-0.21,0-0.37-0.06-0.49-0.18c-0.13-0.12-0.19-0.27-0.19-0.47v-0.27c-0.16,0.3-0.39,0.54-0.7,0.7c-0.3,0.17-0.65,0.25-1.04,0.25
c-0.46,0-0.87-0.12-1.24-0.35c-0.36-0.23-0.65-0.57-0.85-1c-0.2-0.43-0.3-0.92-0.3-1.48c0-0.56,0.1-1.05,0.3-1.46
c0.2-0.42,0.48-0.74,0.84-0.97s0.77-0.34,1.24-0.34c0.39,0,0.73,0.08,1.03,0.24c0.3,0.16,0.53,0.39,0.69,0.68v-2.61
c0-0.18,0.06-0.33,0.18-0.45c0.12-0.11,0.28-0.17,0.49-0.17S393.5,295.18,393.63,295.3z M392.11,301.5
c0.24-0.3,0.36-0.73,0.36-1.29c0-0.56-0.12-0.98-0.36-1.28s-0.57-0.45-1.01-0.45c-0.44,0-0.78,0.15-1.02,0.44s-0.36,0.71-0.36,1.27
c0,0.56,0.12,0.99,0.36,1.3s0.58,0.46,1.01,0.46C391.53,301.95,391.87,301.8,392.11,301.5z" />
</g>
</svg>
<!-- end of perceptronAnimation -->
</div>
<!-- svg goo filter DO NOT DELETE-->
<!-- <svg version="1.1" xmlns="https://www.w3.org/2000/svg" style="height: 0; width: 0; position: absolute; top: -9999px; left: -99990px">
<defs>
<filter color-interpolation-filters="sRGB" id="goo2">
<feGaussianBlur in="SourceGraphic" stdDeviation="8" result="blur" />
<feColorMatrix in="blur" mode="matrix" values="1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 21 -9" result="cm" />
<feBlend />
</filter>
</defs>
</svg> -->
<!-- Boostrap JavaScript -->
<script src="https://code.jquery.com/jquery-3.3.1.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/popper.js/1.14.0/umd/popper.min.js"
integrity="sha384-cs/chFZiN24E4KMATLdqdvsezGxaGsi4hLGOzlXwp5UZB1LY//20VyM2taTB4QvJ"
crossorigin="anonymous"></script>
<script src="https://stackpath.bootstrapcdn.com/bootstrap/4.1.0/js/bootstrap.min.js" integrity="sha384-uefMccjFJAIv6A+rW+L4AHf99KvxDjWSu1z9VI8SKNVmz4sk7buKt/6v9KI65qnm"
crossorigin="anonymous"></script>
<!-- jQuery Custom Scroller CDN -->
<script src="https://cdnjs.cloudflare.com/ajax/libs/malihu-custom-scrollbar-plugin/3.1.5/jquery.mCustomScrollbar.concat.min.js"></script>
<!-- GSAP -->
<script type="text/javascript" src="./js/TweenMax.min.js"></script>
<script type="text/javascript" src="./js/DrawSVGPlugin.min.js"></script>
<script type="text/javascript" src="./js/MorphSVGPlugin.min.js"></script>
<!-- Lottie -->
<script type="text/javascript" src="./js/lottie.js"></script>
<!-- ScrollMagic JavaScript -->
<script src="https://cdnjs.cloudflare.com/ajax/libs/ScrollMagic/2.0.6/ScrollMagic.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/ScrollMagic/2.0.5/plugins/animation.gsap.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/ScrollMagic/2.0.5//plugins/debug.addIndicators.js"></script>
<script type="text/javascript" src="./js/main.js"></script>
<script type="text/javascript" src="./js/chapter3.js"></script>
<script type="text/javascript">
$(document).ready(function() {
$('[data-toggle="tooltip"]').tooltip({
placement: "top"
});
});
</script>
</body>
</html>