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
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
|
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<link rel="canonical" href="http://keras.io/examples/mnist_net2net/">
<link rel="shortcut icon" href="../../img/favicon.ico">
<title>Mnist net2net - Keras Documentation</title>
<link href='https://fonts.googleapis.com/css?family=Lato:400,700|Source+Sans+Pro:400,700|Inconsolata:400,700' rel='stylesheet' type='text/css'>
<link rel="stylesheet" href="../../css/theme.css" type="text/css" />
<link rel="stylesheet" href="../../css/theme_extra.css" type="text/css" />
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/styles/github.min.css">
<script>
// Current page data
var mkdocs_page_name = "Mnist net2net";
var mkdocs_page_input_path = "examples/mnist_net2net.md";
var mkdocs_page_url = "/examples/mnist_net2net/";
</script>
<script src="../../js/jquery-2.1.1.min.js" defer></script>
<script src="../../js/modernizr-2.8.3.min.js" defer></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/highlight.min.js"></script>
<script>hljs.initHighlightingOnLoad();</script>
<script>
(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
})(window,document,'script','https://www.google-analytics.com/analytics.js','ga');
ga('create', 'UA-61785484-1', 'keras.io');
ga('send', 'pageview');
</script>
</head>
<body class="wy-body-for-nav" role="document">
<div class="wy-grid-for-nav">
<nav data-toggle="wy-nav-shift" class="wy-nav-side stickynav">
<div class="wy-side-scroll">
<a href="">
<div class="keras-logo">
<img src="/img/keras-logo-small.jpg" class="keras-logo-img">
Keras Documentation
</div>
</a>
<div class="wy-side-nav-search">
<div role="search">
<form id ="rtd-search-form" class="wy-form" action="../../search.html" method="get">
<input type="text" name="q" placeholder="Search docs" title="Type search term here" />
</form>
</div>
</div>
<div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
<ul>
<li class="toctree-l1"><a class="reference internal" href="../..">Home</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../why-use-keras/">Why use Keras</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Getting started</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../getting-started/sequential-model-guide/">Guide to the Sequential model</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../getting-started/functional-api-guide/">Guide to the Functional API</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../getting-started/faq/">FAQ</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Models</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../models/about-keras-models/">About Keras models</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../models/sequential/">Sequential</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../models/model/">Model (functional API)</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Layers</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../layers/about-keras-layers/">About Keras layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/core/">Core Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/convolutional/">Convolutional Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/pooling/">Pooling Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/local/">Locally-connected Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/recurrent/">Recurrent Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/embeddings/">Embedding Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/merge/">Merge Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/advanced-activations/">Advanced Activations Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/normalization/">Normalization Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/noise/">Noise layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/wrappers/">Layer wrappers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../layers/writing-your-own-keras-layers/">Writing your own Keras layers</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Preprocessing</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../preprocessing/sequence/">Sequence Preprocessing</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../preprocessing/text/">Text Preprocessing</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../preprocessing/image/">Image Preprocessing</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../losses/">Losses</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../metrics/">Metrics</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../optimizers/">Optimizers</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../activations/">Activations</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../callbacks/">Callbacks</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../datasets/">Datasets</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../applications/">Applications</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../backend/">Backend</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../initializers/">Initializers</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../regularizers/">Regularizers</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../constraints/">Constraints</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../visualization/">Visualization</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../scikit-learn-api/">Scikit-learn API</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../utils/">Utils</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../contributing/">Contributing</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Examples</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../addition_rnn/">Addition RNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../antirectifier/">Custom layer - antirectifier</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../babi_rnn/">Baby RNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../babi_memnn/">Baby MemNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../cifar10_cnn/">CIFAR-10 CNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../cifar10_resnet/">CIFAR-10 ResNet</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../conv_filter_visualization/">Convolution filter visualization</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../conv_lstm/">Convolutional LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../deep_dream/">Deep Dream</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../image_ocr/">Image OCR</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../imdb_bidirectional_lstm/">Bidirectional LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../imdb_cnn/">1D CNN for text classification</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../imdb_cnn_lstm/">Sentiment classification CNN-LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../imdb_fasttext/">Fasttext for text classification</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../imdb_lstm/">Sentiment classification LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../lstm_seq2seq/">Sequence to sequence - training</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../lstm_seq2seq_restore/">Sequence to sequence - prediction</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../lstm_stateful/">Stateful LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../lstm_text_generation/">LSTM for text generation</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../mnist_acgan/">Auxiliary Classifier GAN</a>
</li>
</ul>
</div>
</div>
</nav>
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">
<nav class="wy-nav-top" role="navigation" aria-label="top navigation">
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
<a href="../..">Keras Documentation</a>
</nav>
<div class="wy-nav-content">
<div class="rst-content">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="wy-breadcrumbs">
<li><a href="../..">Docs</a> »</li>
<li>Mnist net2net</li>
<li class="wy-breadcrumbs-aside">
<a href="https://github.com/keras-team/keras/tree/master/docs"
class="icon icon-github"> Edit on GitHub</a>
</li>
</ul>
<hr/>
</div>
<div role="main">
<div class="section">
<p>This is an implementation of Net2Net experiment with MNIST in
'Net2Net: Accelerating Learning via Knowledge Transfer'
by Tianqi Chen, Ian Goodfellow, and Jonathon Shlens</p>
<p>arXiv:1511.05641v4 [cs.LG] 23 Apr 2016
http://arxiv.org/abs/1511.05641</p>
<h1 id="notes">Notes</h1>
<ul>
<li>What:</li>
<li>Net2Net is a group of methods to transfer knowledge from a teacher neural
net to a student net,so that the student net can be trained faster than
from scratch.</li>
<li>The paper discussed two specific methods of Net2Net, i.e. Net2WiderNet
and Net2DeeperNet.</li>
<li>Net2WiderNet replaces a model with an equivalent wider model that has
more units in each hidden layer.</li>
<li>Net2DeeperNet replaces a model with an equivalent deeper model.</li>
<li>Both are based on the idea of 'function-preserving transformations of
neural nets'.</li>
<li>Why:</li>
<li>Enable fast exploration of multiple neural nets in experimentation and
design process,by creating a series of wider and deeper models with
transferable knowledge.</li>
<li>Enable 'lifelong learning system' by gradually adjusting model complexity
to data availability,and reusing transferable knowledge.</li>
</ul>
<h1 id="experiments">Experiments</h1>
<ul>
<li>Teacher model: a basic CNN model trained on MNIST for 3 epochs.</li>
<li>Net2WiderNet experiment:</li>
<li>Student model has a wider Conv2D layer and a wider FC layer.</li>
<li>Comparison of 'random-padding' vs 'net2wider' weight initialization.</li>
<li>With both methods, after 1 epoch, student model should perform as well as
teacher model, but 'net2wider' is slightly better.</li>
<li>Net2DeeperNet experiment:</li>
<li>Student model has an extra Conv2D layer and an extra FC layer.</li>
<li>Comparison of 'random-init' vs 'net2deeper' weight initialization.</li>
<li>After 1 epoch, performance of 'net2deeper' is better than 'random-init'.</li>
<li>Hyper-parameters:</li>
<li>SGD with momentum=0.9 is used for training teacher and student models.</li>
<li>Learning rate adjustment: it's suggested to reduce learning rate
to 1/10 for student model.</li>
<li>Addition of noise in 'net2wider' is used to break weight symmetry
and thus enable full capacity of student models. It is optional
when a Dropout layer is used.</li>
</ul>
<h1 id="results">Results</h1>
<ul>
<li>Tested with TF backend and 'channels_last' image_data_format.</li>
<li>Running on GPU GeForce GTX Titan X Maxwell</li>
<li>Performance Comparisons - validation loss values during first 3 epochs:</li>
</ul>
<p>Teacher model ...
(0) teacher_model: 0.0537 0.0354 0.0356</p>
<p>Experiment of Net2WiderNet ...
(1) wider_random_pad: 0.0320 0.0317 0.0289
(2) wider_net2wider: 0.0271 0.0274 0.0270</p>
<p>Experiment of Net2DeeperNet ...
(3) deeper_random_init: 0.0682 0.0506 0.0468
(4) deeper_net2deeper: 0.0292 0.0294 0.0286</p>
<pre><code class="python">from __future__ import print_function
import numpy as np
import keras
from keras import backend as K
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Dense, Flatten
from keras.optimizers import SGD
from keras.datasets import mnist
if K.image_data_format() == 'channels_first':
input_shape = (1, 28, 28) # image shape
else:
input_shape = (28, 28, 1) # image shape
num_classes = 10 # number of classes
epochs = 3
# load and pre-process data
def preprocess_input(x):
return x.astype('float32').reshape((-1,) + input_shape) / 255
def preprocess_output(y):
return keras.utils.to_categorical(y)
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = map(preprocess_input, [x_train, x_test])
y_train, y_test = map(preprocess_output, [y_train, y_test])
print('Loading MNIST data...')
print('x_train shape:', x_train.shape, 'y_train shape:', y_train.shape)
print('x_test shape:', x_test.shape, 'y_test shape', y_test.shape)
# knowledge transfer algorithms
def wider2net_conv2d(teacher_w1, teacher_b1, teacher_w2, new_width, init):
'''Get initial weights for a wider conv2d layer with a bigger filters,
by 'random-padding' or 'net2wider'.
# Arguments
teacher_w1: `weight` of conv2d layer to become wider,
of shape (filters1, num_channel1, kh1, kw1)
teacher_b1: `bias` of conv2d layer to become wider,
of shape (filters1, )
teacher_w2: `weight` of next connected conv2d layer,
of shape (filters2, num_channel2, kh2, kw2)
new_width: new `filters` for the wider conv2d layer
init: initialization algorithm for new weights,
either 'random-pad' or 'net2wider'
'''
assert teacher_w1.shape[0] == teacher_w2.shape[1], (
'successive layers from teacher model should have compatible shapes')
assert teacher_w1.shape[3] == teacher_b1.shape[0], (
'weight and bias from same layer should have compatible shapes')
assert new_width > teacher_w1.shape[3], (
'new width (filters) should be bigger than the existing one')
n = new_width - teacher_w1.shape[3]
if init == 'random-pad':
new_w1 = np.random.normal(0, 0.1, size=teacher_w1.shape[:3] + (n,))
new_b1 = np.ones(n) * 0.1
new_w2 = np.random.normal(
0, 0.1,
size=teacher_w2.shape[:2] + (n, teacher_w2.shape[3]))
elif init == 'net2wider':
index = np.random.randint(teacher_w1.shape[3], size=n)
factors = np.bincount(index)[index] + 1.
new_w1 = teacher_w1[:, :, :, index]
new_b1 = teacher_b1[index]
new_w2 = teacher_w2[:, :, index, :] / factors.reshape((1, 1, -1, 1))
else:
raise ValueError('Unsupported weight initializer: %s' % init)
student_w1 = np.concatenate((teacher_w1, new_w1), axis=3)
if init == 'random-pad':
student_w2 = np.concatenate((teacher_w2, new_w2), axis=2)
elif init == 'net2wider':
# add small noise to break symmetry, so that student model will have
# full capacity later
noise = np.random.normal(0, 5e-2 * new_w2.std(), size=new_w2.shape)
student_w2 = np.concatenate((teacher_w2, new_w2 + noise), axis=2)
student_w2[:, :, index, :] = new_w2
student_b1 = np.concatenate((teacher_b1, new_b1), axis=0)
return student_w1, student_b1, student_w2
def wider2net_fc(teacher_w1, teacher_b1, teacher_w2, new_width, init):
'''Get initial weights for a wider fully connected (dense) layer
with a bigger nout, by 'random-padding' or 'net2wider'.
# Arguments
teacher_w1: `weight` of fc layer to become wider,
of shape (nin1, nout1)
teacher_b1: `bias` of fc layer to become wider,
of shape (nout1, )
teacher_w2: `weight` of next connected fc layer,
of shape (nin2, nout2)
new_width: new `nout` for the wider fc layer
init: initialization algorithm for new weights,
either 'random-pad' or 'net2wider'
'''
assert teacher_w1.shape[1] == teacher_w2.shape[0], (
'successive layers from teacher model should have compatible shapes')
assert teacher_w1.shape[1] == teacher_b1.shape[0], (
'weight and bias from same layer should have compatible shapes')
assert new_width > teacher_w1.shape[1], (
'new width (nout) should be bigger than the existing one')
n = new_width - teacher_w1.shape[1]
if init == 'random-pad':
new_w1 = np.random.normal(0, 0.1, size=(teacher_w1.shape[0], n))
new_b1 = np.ones(n) * 0.1
new_w2 = np.random.normal(0, 0.1, size=(n, teacher_w2.shape[1]))
elif init == 'net2wider':
index = np.random.randint(teacher_w1.shape[1], size=n)
factors = np.bincount(index)[index] + 1.
new_w1 = teacher_w1[:, index]
new_b1 = teacher_b1[index]
new_w2 = teacher_w2[index, :] / factors[:, np.newaxis]
else:
raise ValueError('Unsupported weight initializer: %s' % init)
student_w1 = np.concatenate((teacher_w1, new_w1), axis=1)
if init == 'random-pad':
student_w2 = np.concatenate((teacher_w2, new_w2), axis=0)
elif init == 'net2wider':
# add small noise to break symmetry, so that student model will have
# full capacity later
noise = np.random.normal(0, 5e-2 * new_w2.std(), size=new_w2.shape)
student_w2 = np.concatenate((teacher_w2, new_w2 + noise), axis=0)
student_w2[index, :] = new_w2
student_b1 = np.concatenate((teacher_b1, new_b1), axis=0)
return student_w1, student_b1, student_w2
def deeper2net_conv2d(teacher_w):
'''Get initial weights for a deeper conv2d layer by net2deeper'.
# Arguments
teacher_w: `weight` of previous conv2d layer,
of shape (kh, kw, num_channel, filters)
'''
kh, kw, num_channel, filters = teacher_w.shape
student_w = np.zeros_like(teacher_w)
for i in range(filters):
student_w[(kh - 1) // 2, (kw - 1) // 2, i, i] = 1.
student_b = np.zeros(filters)
return student_w, student_b
def copy_weights(teacher_model, student_model, layer_names):
'''Copy weights from teacher_model to student_model,
for layers with names listed in layer_names
'''
for name in layer_names:
weights = teacher_model.get_layer(name=name).get_weights()
student_model.get_layer(name=name).set_weights(weights)
# methods to construct teacher_model and student_models
def make_teacher_model(x_train, y_train,
x_test, y_test,
epochs):
'''Train and benchmark performance of a simple CNN.
(0) Teacher model
'''
model = Sequential()
model.add(Conv2D(64, 3, input_shape=input_shape,
padding='same', name='conv1'))
model.add(MaxPooling2D(2, name='pool1'))
model.add(Conv2D(64, 3, padding='same', name='conv2'))
model.add(MaxPooling2D(2, name='pool2'))
model.add(Flatten(name='flatten'))
model.add(Dense(64, activation='relu', name='fc1'))
model.add(Dense(num_classes, activation='softmax', name='fc2'))
model.compile(loss='categorical_crossentropy',
optimizer=SGD(learning_rate=0.01, momentum=0.9),
metrics=['accuracy'])
model.fit(x_train, y_train,
epochs=epochs,
validation_data=(x_test, y_test))
return model
def make_wider_student_model(teacher_model,
x_train, y_train,
x_test, y_test,
init, epochs):
'''Train a wider student model based on teacher_model,
with either 'random-pad' (baseline) or 'net2wider'
'''
new_conv1_width = 128
new_fc1_width = 128
model = Sequential()
# a wider conv1 compared to teacher_model
model.add(Conv2D(new_conv1_width, 3, input_shape=input_shape,
padding='same', name='conv1'))
model.add(MaxPooling2D(2, name='pool1'))
model.add(Conv2D(64, 3, padding='same', name='conv2'))
model.add(MaxPooling2D(2, name='pool2'))
model.add(Flatten(name='flatten'))
# a wider fc1 compared to teacher model
model.add(Dense(new_fc1_width, activation='relu', name='fc1'))
model.add(Dense(num_classes, activation='softmax', name='fc2'))
# The weights for other layers need to be copied from teacher_model
# to student_model, except for widened layers
# and their immediate downstreams, which will be initialized separately.
# For this example there are no other layers that need to be copied.
w_conv1, b_conv1 = teacher_model.get_layer('conv1').get_weights()
w_conv2, b_conv2 = teacher_model.get_layer('conv2').get_weights()
new_w_conv1, new_b_conv1, new_w_conv2 = wider2net_conv2d(
w_conv1, b_conv1, w_conv2, new_conv1_width, init)
model.get_layer('conv1').set_weights([new_w_conv1, new_b_conv1])
model.get_layer('conv2').set_weights([new_w_conv2, b_conv2])
w_fc1, b_fc1 = teacher_model.get_layer('fc1').get_weights()
w_fc2, b_fc2 = teacher_model.get_layer('fc2').get_weights()
new_w_fc1, new_b_fc1, new_w_fc2 = wider2net_fc(
w_fc1, b_fc1, w_fc2, new_fc1_width, init)
model.get_layer('fc1').set_weights([new_w_fc1, new_b_fc1])
model.get_layer('fc2').set_weights([new_w_fc2, b_fc2])
model.compile(loss='categorical_crossentropy',
optimizer=SGD(learning_rate=0.001, momentum=0.9),
metrics=['accuracy'])
model.fit(x_train, y_train,
epochs=epochs,
validation_data=(x_test, y_test))
def make_deeper_student_model(teacher_model,
x_train, y_train,
x_test, y_test,
init, epochs):
'''Train a deeper student model based on teacher_model,
with either 'random-init' (baseline) or 'net2deeper'
'''
model = Sequential()
model.add(Conv2D(64, 3, input_shape=input_shape,
padding='same', name='conv1'))
model.add(MaxPooling2D(2, name='pool1'))
model.add(Conv2D(64, 3, padding='same', name='conv2'))
# add another conv2d layer to make original conv2 deeper
if init == 'net2deeper':
prev_w, _ = model.get_layer('conv2').get_weights()
new_weights = deeper2net_conv2d(prev_w)
model.add(Conv2D(64, 3, padding='same',
name='conv2-deeper', weights=new_weights))
elif init == 'random-init':
model.add(Conv2D(64, 3, padding='same', name='conv2-deeper'))
else:
raise ValueError('Unsupported weight initializer: %s' % init)
model.add(MaxPooling2D(2, name='pool2'))
model.add(Flatten(name='flatten'))
model.add(Dense(64, activation='relu', name='fc1'))
# add another fc layer to make original fc1 deeper
if init == 'net2deeper':
# net2deeper for fc layer with relu, is just an identity initializer
model.add(Dense(64, kernel_initializer='identity',
activation='relu', name='fc1-deeper'))
elif init == 'random-init':
model.add(Dense(64, activation='relu', name='fc1-deeper'))
else:
raise ValueError('Unsupported weight initializer: %s' % init)
model.add(Dense(num_classes, activation='softmax', name='fc2'))
# copy weights for other layers
copy_weights(teacher_model, model, layer_names=[
'conv1', 'conv2', 'fc1', 'fc2'])
model.compile(loss='categorical_crossentropy',
optimizer=SGD(learning_rate=0.001, momentum=0.9),
metrics=['accuracy'])
model.fit(x_train, y_train,
epochs=epochs,
validation_data=(x_test, y_test))
# experiments setup
def net2wider_experiment():
'''Benchmark performances of
(1) a wider student model with `random_pad` initializer
(2) a wider student model with `Net2WiderNet` initializer
'''
print('\nExperiment of Net2WiderNet ...')
print('\n(1) building wider student model by random padding ...')
make_wider_student_model(teacher_model,
x_train, y_train,
x_test, y_test,
init='random-pad',
epochs=epochs)
print('\n(2) building wider student model by net2wider ...')
make_wider_student_model(teacher_model,
x_train, y_train,
x_test, y_test,
init='net2wider',
epochs=epochs)
def net2deeper_experiment():
'''Benchmark performances of
(3) a deeper student model with `random_init` initializer
(4) a deeper student model with `Net2DeeperNet` initializer
'''
print('\nExperiment of Net2DeeperNet ...')
print('\n(3) building deeper student model by random init ...')
make_deeper_student_model(teacher_model,
x_train, y_train,
x_test, y_test,
init='random-init',
epochs=epochs)
print('\n(4) building deeper student model by net2deeper ...')
make_deeper_student_model(teacher_model,
x_train, y_train,
x_test, y_test,
init='net2deeper',
epochs=epochs)
print('\n(0) building teacher model ...')
teacher_model = make_teacher_model(x_train, y_train,
x_test, y_test,
epochs=epochs)
# run the experiments
net2wider_experiment()
net2deeper_experiment()
</code></pre>
</div>
</div>
<footer>
<hr/>
<div role="contentinfo">
<!-- Copyright etc -->
</div>
Built with <a href="https://www.mkdocs.org/">MkDocs</a> using a <a href="https://github.com/snide/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</footer>
</div>
</div>
</section>
</div>
<div class="rst-versions" role="note" aria-label="versions">
<span class="rst-current-version" data-toggle="rst-current-version">
<a href="http://github.com/keras-team/keras/" class="fa fa-github" style="float: left; color: #fcfcfc"> GitHub</a>
</span>
</div>
<script>var base_url = '../..';</script>
<script src="../../js/theme.js" defer></script>
<script src="../../search/main.js" defer></script>
<script type="text/javascript" defer>
window.onload = function () {
SphinxRtdTheme.Navigation.enable(true);
};
</script>
</body>
</html>
|
