forked from cs231n/cs231n.github.io
-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathsyllabus.html
475 lines (459 loc) · 18 KB
/
syllabus.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
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1">
<!-- The above 3 meta tags *must* come first in the head; any other head content must come *after* these tags -->
<meta name="description" content="">
<meta name="author" content="">
<!-- HTML4 meta tags forcing no-caching -->
<meta http-equiv="Cache-Control" content="no-cache, no-store, must-revalidate" />
<meta http-equiv="Pragma" content="no-cache" />
<meta http-equiv="Expires" content="0" />
<link rel="icon" href="/favicon.ico?">
<title>COMPSCI 697L Deep Learning</title>
<!-- Bootstrap core CSS -->
<link href="assets/css/bootstrap.min.css" rel="stylesheet">
<!-- IE10 viewport hack for Surface/desktop Windows 8 bug -->
<link href="assets/css/ie10-viewport-bug-workaround.css" rel="stylesheet">
<!-- Custom styles for this template -->
<link href="assets/css/offcanvas.css" rel="stylesheet">
<!-- Just for debugging purposes. Don't actually copy these 2 lines! -->
<!--[if lt IE 9]><script src="../../assets/js/ie8-responsive-file-warning.js"></script><![endif]-->
<script src="assets/js/ie-emulation-modes-warning.js"></script>
<!-- HTML5 shim and Respond.js for IE8 support of HTML5 elements and media queries -->
<!--[if lt IE 9]>
<script src="https://oss.maxcdn.com/html5shiv/3.7.3/html5shiv.min.js"></script>
<script src="https://oss.maxcdn.com/respond/1.4.2/respond.min.js"></script>
<![endif]-->
</head>
<body>
<nav class="navbar navbar-fixed-top navbar-custom">
<div class="container">
<div class="navbar-header">
<button type="button" class="navbar-toggle collapsed" data-toggle="collapse" data-target="#navbar" aria-expanded="false" aria-controls="navbar">
<span class="sr-only">Toggle navigation</span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
</button>
<!--<a class="navbar-brand" href="https://www.cics.umass.edu/">COMPSCI697L</a>-->
<a class="navbar-brand" href="https://www.cics.umass.edu/"><img style="max-height:15px; margin-top: 1px;"
src="/assets/fig/umasslogo2.png"></a>
<!--<a class="navbar-brand" href="#">COMPSCI697L</a>-->
</div>
<div id="navbar" class="collapse navbar-collapse">
<ul class="nav navbar-nav">
<li><a href="/index.html">Home</a></li>
<li class="active"><a href="/syllabus.html">Syllabus</a></li>
<li><a href="/notes/">Notes</a></li>
<li><a href="/assignments.html">Assignments</a></li>
<li><a href="/projects/">Project</a></li>
</ul>
</div><!-- /.nav-collapse -->
</div><!-- /.container -->
</nav><!-- /.navbar -->
<div class="container">
<h2>COMPSCI 697L Deep Learning</h2>
<div class="panel panel-info">
<div class="panel-heading">
<h3 class="panel-title">Note</h3>
</div>
<div class="panel-body">
<ul>
<li>This is a tentative class outline and is subject to change throughout the semester. </li>
<li>Slides will be posted after each lecture.</li>
</ul>
</div>
</div>
<div class="panel panel-default">
<table class="table">
<tbody><tr class="active">
<th>Event Type</th><th>Date</th><th>Description</th><th>Course Materials</th>
</tr>
<tr>
<td>Lecture</td>
<td>Sep 7</td>
<td>Intro to Deep Learning, historical context.</td>
<td>
<a href="https://docs.google.com/presentation/d/1l6VCpuOA1AlJAjZ-6tsGRd-YTojxMaRVPCYI2sWe3Dg/edit?usp=sharing">[slides]</a><br>
<a href="notes/python-numpy-tutorial/">[python/numpy tutorial]</a><br>
<a href="notes/jupyter-tutorial/">[jupyter tutorial]</a>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Sep 12</td>
<td>Image classification and the data-driven approach <br> k-nearest neighbor <br> Linear classification I</td>
<td>
<a href="https://docs.google.com/presentation/d/10vfD1Q9wfZcVUU_uG5uhdnbnk5nocRz__B5AoHxWNlI/edit?usp=sharing">[slides]</a><br>
<a href="notes/classification/">[image classification notes]</a><br>
<a href="notes/linear-classify/">[linear classification notes]</a><br>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Sep 14</td>
<td>
Linear classification II<br>
</td>
<td>
<a href="https://docs.google.com/presentation/d/1JRqOoslZ1T8cIqBxL8oHoz2F9wZ_v3al0yWpWmyyo4w/edit?usp=sharing">[slides]</a><br>
<a href="notes/linear-classify/">[linear classification notes]</a>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Sep 19</td>
<td>
Loss functions<br>
Optimization I: Stochastic gradient descent
</td>
<td>
<a href="https://docs.google.com/presentation/d/1kQmh9T8t7n2ogJuLtVKWNswe24wdxHALvPDUK6j1abc/edit?usp=sharing">[slides]</a><br>
<a href="notes/optimization-1/">[optimization notes]</a>
</td>
</tr>
<tr>
<tr>
<td>Lecture</td>
<td>Sep 21</td>
<td>
Backpropagation & Neural Networks I
</td>
<td>
<a href="https://docs.google.com/presentation/d/1oudCntxqxQpH-yPmPhqEvmyqlaa8EahsMjbhvwg4W_U/edit?usp=sharing">[slides]</a><br>
<a href="notes/optimization-2/">[backprop notes]</a><br>
<a href="http://yann.lecun.com/exdb/publis/pdf/lecun-98b.pdf">[Efficient BackProp]</a> (optional)<br>
related: <a href="http://colah.github.io/posts/2015-08-Backprop/">[1]</a>, <a href="http://neuralnetworksanddeeplearning.com/chap2.html">[2]</a>, <a href="https://www.youtube.com/watch?v=q0pm3BrIUFo">[3]</a> (optional)
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Sep 26</td>
<td>
Higher-level representations, image features<br>
Vector, Matrix, and Tensor Derivatives
</td>
<td>
<a href="/docs/vecDerivs.pdf">[handout]</a><br>
<a href="https://docs.google.com/presentation/d/1h_7jVx-75vfs6m5qbyr4T1M13bbya_K505_jEfXRpAA/edit?usp=sharing">[slides]</a><br>
<a href="http://www.nature.com/nature/journal/v521/n7553/full/nature14539.html">Deep Learning [Nature]</a> (optional)
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Sep 28</td>
<td>
Backpropagation & Neural Networks II
</td>
<td>
<a href="https://docs.google.com/presentation/d/1h_7jVx-75vfs6m5qbyr4T1M13bbya_K505_jEfXRpAA/edit?usp=sharing">[slides]</a> (cont.)<br>
tips/tricks:
<a href="http://research.microsoft.com/pubs/192769/tricks-2012.pdf">[1]</a>,
<a href="http://arxiv.org/pdf/1206.5533v2.pdf">[2]</a> (optional)
<br>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 3</td>
<td>
Training Neural Networks I
</td>
<td>
<a href="https://docs.google.com/presentation/d/1-EIQUEk7unzawJ-iK0PmpQaveLK2uLdc7EDF6wUJOvA/edit?usp=sharing">[slides]</a><br>
<a href="notes/neural-networks-1/">[Neural Nets notes 1]</a>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 5</td>
<td>
Training Neural Networks II
</td>
<td>
<a href="https://docs.google.com/presentation/d/1V2lgXXXFs_1qRe77XP9PUzvEUCrZ1L5BRAQc4CRZoz8/edit?usp=sharing">[slides]</a><br>
<a href="http://yann.lecun.com/exdb/publis/pdf/lecun-98.pdf">LeNet</a> (optional)
</td>
</tr>
<tr class="info">
<td>No class</td>
<td>Oct 10</td>
<td>Columbus Day; Class will be on Tuesday (Oct 11) instead.</td>
<td></td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 11</td>
<td>
Training Neural Networks III: <br>
weight initialization, batch normalization
</td>
<td>
<a href="https://docs.google.com/presentation/d/1UhJiujgYX1Z51tFtLprhYGOZZRoKB8Fur91NT580Cvk/edit?usp=sharing">[slides]</a><br>
<a href="notes/neural-networks-2/">[Neural Nets notes 2]</a><br>
<a href="https://arxiv.org/abs/1502.03167">[Batch Norm]</a><br>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 12</td>
<td>
Training Neural Network IV: <br>
babysitting the learning process, hyperparameter optimization
</td>
<td>
<a href="https://docs.google.com/presentation/d/1yCF-mHZpPdCOh5YHp0m5YoyrQLzIcF_iymTXLEjiLCk/edit?usp=sharing">[slides]</a><br>
<a href="notes/neural-networks-3/">[Neural Nets notes 3]</a>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 17</td>
<td>
Training Neural Network IV (cont.): <br>
babysitting the learning process, hyperparameter optimization
</td>
<td>
<a href="https://docs.google.com/presentation/d/1yCF-mHZpPdCOh5YHp0m5YoyrQLzIcF_iymTXLEjiLCk/edit?usp=sharing">[slides]</a> (cont.)<br>
<a href="https://arxiv.org/abs/1206.5533">[Bengio 2012]</a> (optional)<br>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 19</td>
<td>
Project announcement <br>
Training Neural Network V: <br>
parameter updates, model ensembles, dropout
</td>
<td>
<a href="https://docs.google.com/presentation/d/1wA0TwJ_Gd4rIwelljTvcVdc5gCUi5taKBPl2PZkUXME/edit?usp=sharing">[slides]</a><br>
<a href="http://cs231n.stanford.edu/reports.html">[Stanford cs231n project reports]</a>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 24</td>
<td>
Convolutional Neural Networks: introduction, history, architectures
</td>
<td>
<a href="https://docs.google.com/presentation/d/1mijEyEIKPj8rhfb_CLLs45ZAVGzx_c-7mMiDynCiSE4/edit?usp=sharing">[slides]</a><br>
<a href="notes/convolutional-networks/">[ConvNet notes]</a> <br>
<a href="https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf">AlexNet</a> (optional)
</td>
</tr>
<tr>
<td>Guest Lecture</td>
<td>Oct 26</td>
<td> <a href="https://people.cs.umass.edu/~tsungyulin/">Tsung-Yu Lin</a>: Bilinear CNN
</td>
<td>
<a href="https://drive.google.com/file/d/0B-0OtUj_Gj7nOHVNSjJ3OVFCdUU/view?usp=sharing">[slides]</a><br>
<a href="https://arxiv.org/abs/1504.07889">BCNN</a> (optional)
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Oct 31</td>
<td>
Convolutional Neural Networks: convolution layer, pooling layer, fully connected layer <br>
Case study of ImageNet challenge winning ConvNets
</td>
<td>
<a href="https://docs.google.com/presentation/d/1mijEyEIKPj8rhfb_CLLs45ZAVGzx_c-7mMiDynCiSE4/edit?usp=sharing">[slides]</a> (cont.)<br>
<a href="https://drive.google.com/file/d/0B-0OtUj_Gj7namdSdDZsVXdnLTl6YWlJaTdkODc2eWZMT1Nv/view?usp=sharing">[midterm review sheet]</a>
</td>
</tr>
<tr class="danger">
<td>Midterm</td>
<td>Nov 2</td>
<td>
In-class midterm<br>
Project proposals due!
</td>
<td>
<a href="https://docs.google.com/spreadsheets/d/1X8A8FrB0T6zfT6-xekzqrVK3h67jf78mUQUP52ogxKw/edit?usp=sharing">Proposed project topics</a>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Nov 7</td>
<td>
Case study of ImageNet challenge winning ConvNets (cont.) <br>
ConvNets for spatial localization, Object detection
</td>
<td>
<a href="https://docs.google.com/presentation/d/17P1pFTG7seo6EuDhLrJDI6Bb7ce1_mUd28qhJKe5j_k/edit?usp=sharing">[slides]</a><br>
<a href="https://arxiv.org/abs/1512.03385">ResNet</a> (optional)
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Nov 9</td>
<td>
ConvNets for spatial localization, Object detection (cont.)
</td>
<td>
<a href="https://docs.google.com/presentation/d/17P1pFTG7seo6EuDhLrJDI6Bb7ce1_mUd28qhJKe5j_k/edit?usp=sharing">[slides]</a> (cont.)<br>
<a href="http://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Long_Fully_Convolutional_Networks_2015_CVPR_paper.pdf">FCN</a> (optional)
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Nov 14</td>
<td>
Understanding and visualizing Convolutional Neural Networks<br>
Backprop into image: Visualizations, deep dream
</td>
<td>
<a href="https://docs.google.com/presentation/d/16VDgX6BwCkJvsJOxKAcjFFbnfWb3A2069gyGOwHWSXY/edit?usp=sharing">[slides]</a><br>
<a href="notes/understanding-cnn/">[visualization notes]</a>
</td>
</tr>
<tr class="info">
<td>No class</td>
<td>Nov 16</td>
<td>
Friday schedule; No class.
</td>
<td>
</td>
</tr>
<tr class="info">
<td>No class</td>
<td>Nov 21</td>
<td>
Thanksgiving; No class.
</td>
<td>
</td>
</tr>
<tr class="info">
<td>No class</td>
<td>Nov 23</td>
<td>
Thanksgiving; No class.
</td>
<td>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Nov 28</td>
<td>
Artistic style transfer<br>
Adversarial fooling examples<br>
Recurrent Neural Networks (RNN)
</td>
<td>
<a href="https://docs.google.com/presentation/d/1tv3BCnd8hGk4O_1Rmi4qx2AWinDdYM4dxyS-FMlDlZI/edit?usp=sharing">[slides]</a><br>
<a href="http://www.deeplearningbook.org/contents/rnn.html">DL book RNN chapter</a> (optional)<br>
<a href="https://gist.github.com/karpathy/d4dee566867f8291f086">min-char-rnn</a>, <a href="https://github.com/karpathy/char-rnn">char-rnn</a>, <a href="https://github.com/karpathy/neuraltalk2">neuraltalk2</a>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Nov 30</td>
<td>
Recurrent Neural Networks (RNN) (cont.) <br>
Long Short Term Memory (LSTM)
</td>
<td>
<a href="https://docs.google.com/presentation/d/1tv3BCnd8hGk4O_1Rmi4qx2AWinDdYM4dxyS-FMlDlZI/edit?usp=sharing">[slides]</a> (cont.)<br>
<a href="http://karpathy.github.io/2015/05/21/rnn-effectiveness/">The Unreasonable Effectiveness of RNN</a> (optional)
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Dec 5</td>
<td>
Long Short Term Memory (LSTM) (cont.)
</td>
<td>
<a href="https://docs.google.com/presentation/d/1tv3BCnd8hGk4O_1Rmi4qx2AWinDdYM4dxyS-FMlDlZI/edit?usp=sharing">[slides]</a> (cont.)<br>
<a href="http://colah.github.io/posts/2015-08-Understanding-LSTMs/">Understanding LSTM Networks</a> (optional)
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Dec 7</td>
<td>
Training ConvNets in practice
</td>
<td>
<a href="https://docs.google.com/presentation/d/1mMXpk2DVgTqoArOyvm35U-DnW7-X5gCqXZIqhyH4up4/edit?usp=sharing">[slides]</a><br>
</td>
</tr>
<tr>
<td>Lecture</td>
<td>Dec 12</td>
<td>
Additional topics in Stanford cs231n <br>
Societal implications of AI
</td>
<td>
<a href="https://docs.google.com/presentation/d/1qTtKDajhlb5fJfXDDKNGy1C5uqOOvlNSKt4qUfZlseM/edit?usp=sharing">[slides]</a><br>
Stanford cs231n slides not covered in our course: <a href="https://drive.google.com/file/d/0B-0OtUj_Gj7nY2toWTZIekZyRkk/view?usp=sharing">Software Packages</a>, <a href="https://drive.google.com/file/d/0B-0OtUj_Gj7nazc5ajZhZzhKTnM/view?usp=sharing">Segmentation & Attention</a>, <a href="https://drive.google.com/file/d/0B-0OtUj_Gj7nbVRkbm1tUnRUanM/view?usp=sharing">Videos & Unsupervised Learning</a>
</td>
</tr>
<tr class="danger">
<td>Presentation</td>
<td>Dec 14</td>
<td>
Poster presentations at room 150/151
</td>
<td>
<a href="https://docs.google.com/spreadsheets/d/1X8A8FrB0T6zfT6-xekzqrVK3h67jf78mUQUP52ogxKw/edit?usp=sharing">schedule</a>
</td>
</tr>
</tbody></table>
</div>
</div><!--/.container-->
<footer class="site-footer">
<div class="wrap">
<div class="footer-col-1 column">
<ul>
<li><a href="https://github.com/compsci697l">
<span class="icon github">
<svg version="1.1" class="github-icon-svg" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" x="0px" y="0px"
viewBox="0 0 16 16" enable-background="new 0 0 16 16" xml:space="preserve">
<path fill-rule="evenodd" clip-rule="evenodd" fill="#C2C2C2" d="M7.999,0.431c-4.285,0-7.76,3.474-7.76,7.761
c0,3.428,2.223,6.337,5.307,7.363c0.388,0.071,0.53-0.168,0.53-0.374c0-0.184-0.007-0.672-0.01-1.32
c-2.159,0.469-2.614-1.04-2.614-1.04c-0.353-0.896-0.862-1.135-0.862-1.135c-0.705-0.481,0.053-0.472,0.053-0.472
c0.779,0.055,1.189,0.8,1.189,0.8c0.692,1.186,1.816,0.843,2.258,0.645c0.071-0.502,0.271-0.843,0.493-1.037
C4.86,11.425,3.049,10.76,3.049,7.786c0-0.847,0.302-1.54,0.799-2.082C3.768,5.507,3.501,4.718,3.924,3.65
c0,0,0.652-0.209,2.134,0.796C6.677,4.273,7.34,4.187,8,4.184c0.659,0.003,1.323,0.089,1.943,0.261
c1.482-1.004,2.132-0.796,2.132-0.796c0.423,1.068,0.157,1.857,0.077,2.054c0.497,0.542,0.798,1.235,0.798,2.082
c0,2.981-1.814,3.637-3.543,3.829c0.279,0.24,0.527,0.713,0.527,1.437c0,1.037-0.01,1.874-0.01,2.129
c0,0.208,0.14,0.449,0.534,0.373c3.081-1.028,5.302-3.935,5.302-7.362C15.76,3.906,12.285,0.431,7.999,0.431z"/>
</svg>
</span>
<span class="username">compsci697l</span>
</a>
</li>
<li>
<a href="mailto:[email protected]">[email protected]</a>
</li>
</ul>
</div>
<div class="footer-col-2 column">
</div>
<div class="footer-col-3 column">
</div>
</div>
</footer>
<!-- Bootstrap core JavaScript
================================================== -->
<!-- Placed at the end of the document so the pages load faster -->
<script src="https://ajax.googleapis.com/ajax/libs/jquery/1.12.4/jquery.min.js"></script>
<script>window.jQuery || document.write('<script src="../../assets/js/vendor/jquery.min.js"><\/script>')</script>
<script src="assets/js/bootstrap.min.js"></script>
<!-- IE10 viewport hack for Surface/desktop Windows 8 bug -->
<script src="assets/js/ie10-viewport-bug-workaround.js"></script>
<script src="assets/js/offcanvas.js"></script>
</body>
</html>