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
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
|
<!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/layers/core/">
<link rel="shortcut icon" href="../../img/favicon.ico">
<title>Core Layers - 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 = "Core Layers";
var mkdocs_page_input_path = "layers/core.md";
var mkdocs_page_url = "/layers/core/";
</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 class="current">
<li class="toctree-l1"><a class="reference internal" href="../about-keras-layers/">About Keras layers</a>
</li>
<li class="toctree-l1 current"><a class="reference internal current" href="./">Core Layers</a>
<ul class="current">
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../convolutional/">Convolutional Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../pooling/">Pooling Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../local/">Locally-connected Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../recurrent/">Recurrent Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../embeddings/">Embedding Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../merge/">Merge Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../advanced-activations/">Advanced Activations Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../normalization/">Normalization Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../noise/">Noise layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../wrappers/">Layer wrappers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../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="../../examples/addition_rnn/">Addition RNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/antirectifier/">Custom layer - antirectifier</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/babi_rnn/">Baby RNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/babi_memnn/">Baby MemNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/cifar10_cnn/">CIFAR-10 CNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/cifar10_resnet/">CIFAR-10 ResNet</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/conv_filter_visualization/">Convolution filter visualization</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/conv_lstm/">Convolutional LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/deep_dream/">Deep Dream</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/image_ocr/">Image OCR</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/imdb_bidirectional_lstm/">Bidirectional LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/imdb_cnn/">1D CNN for text classification</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/imdb_cnn_lstm/">Sentiment classification CNN-LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/imdb_fasttext/">Fasttext for text classification</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/imdb_lstm/">Sentiment classification LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/lstm_seq2seq/">Sequence to sequence - training</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/lstm_seq2seq_restore/">Sequence to sequence - prediction</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/lstm_stateful/">Stateful LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/lstm_text_generation/">LSTM for text generation</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../examples/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>Layers »</li>
<li>Core Layers</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><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L796">[source]</a></span></p>
<h3 id="dense">Dense</h3>
<pre><code class="python">keras.layers.Dense(units, activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None)
</code></pre>
<p>Just your regular densely-connected NN layer.</p>
<p><code>Dense</code> implements the operation:
<code>output = activation(dot(input, kernel) + bias)</code>
where <code>activation</code> is the element-wise activation function
passed as the <code>activation</code> argument, <code>kernel</code> is a weights matrix
created by the layer, and <code>bias</code> is a bias vector created by the layer
(only applicable if <code>use_bias</code> is <code>True</code>).</p>
<p>Note: if the input to the layer has a rank greater than 2, then
it is flattened prior to the initial dot product with <code>kernel</code>.</p>
<p><strong>Example</strong></p>
<pre><code class="python"># as first layer in a sequential model:
model = Sequential()
model.add(Dense(32, input_shape=(16,)))
# now the model will take as input arrays of shape (*, 16)
# and output arrays of shape (*, 32)
# after the first layer, you don't need to specify
# the size of the input anymore:
model.add(Dense(32))
</code></pre>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>units</strong>: Positive integer, dimensionality of the output space.</li>
<li><strong>activation</strong>: Activation function to use
(see <a href="../../activations/">activations</a>).
If you don't specify anything, no activation is applied
(ie. "linear" activation: <code>a(x) = x</code>).</li>
<li><strong>use_bias</strong>: Boolean, whether the layer uses a bias vector.</li>
<li><strong>kernel_initializer</strong>: Initializer for the <code>kernel</code> weights matrix
(see <a href="../../initializers/">initializers</a>).</li>
<li><strong>bias_initializer</strong>: Initializer for the bias vector
(see <a href="../../initializers/">initializers</a>).</li>
<li><strong>kernel_regularizer</strong>: Regularizer function applied to
the <code>kernel</code> weights matrix
(see <a href="../../regularizers/">regularizer</a>).</li>
<li><strong>bias_regularizer</strong>: Regularizer function applied to the bias vector
(see <a href="../../regularizers/">regularizer</a>).</li>
<li><strong>activity_regularizer</strong>: Regularizer function applied to
the output of the layer (its "activation").
(see <a href="../../regularizers/">regularizer</a>).</li>
<li><strong>kernel_constraint</strong>: Constraint function applied to
the <code>kernel</code> weights matrix
(see <a href="../../constraints/">constraints</a>).</li>
<li><strong>bias_constraint</strong>: Constraint function applied to the bias vector
(see <a href="../../constraints/">constraints</a>).</li>
</ul>
<p><strong>Input shape</strong></p>
<p>nD tensor with shape: <code>(batch_size, ..., input_dim)</code>.
The most common situation would be
a 2D input with shape <code>(batch_size, input_dim)</code>.</p>
<p><strong>Output shape</strong></p>
<p>nD tensor with shape: <code>(batch_size, ..., units)</code>.
For instance, for a 2D input with shape <code>(batch_size, input_dim)</code>,
the output would have shape <code>(batch_size, units)</code>.</p>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L277">[source]</a></span></p>
<h3 id="activation">Activation</h3>
<pre><code class="python">keras.layers.Activation(activation)
</code></pre>
<p>Applies an activation function to an output.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>activation</strong>: name of activation function to use
(see: <a href="../../activations/">activations</a>),
or alternatively, a Theano or TensorFlow operation.</li>
</ul>
<p><strong>Input shape</strong></p>
<p>Arbitrary. Use the keyword argument <code>input_shape</code>
(tuple of integers, does not include the samples axis)
when using this layer as the first layer in a model.</p>
<p><strong>Output shape</strong></p>
<p>Same shape as input.</p>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L81">[source]</a></span></p>
<h3 id="dropout">Dropout</h3>
<pre><code class="python">keras.layers.Dropout(rate, noise_shape=None, seed=None)
</code></pre>
<p>Applies Dropout to the input.</p>
<p>Dropout consists in randomly setting
a fraction <code>rate</code> of input units to 0 at each update during training time,
which helps prevent overfitting.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>rate</strong>: float between 0 and 1. Fraction of the input units to drop.</li>
<li><strong>noise_shape</strong>: 1D integer tensor representing the shape of the
binary dropout mask that will be multiplied with the input.
For instance, if your inputs have shape
<code>(batch_size, timesteps, features)</code> and
you want the dropout mask to be the same for all timesteps,
you can use <code>noise_shape=(batch_size, 1, features)</code>.</li>
<li><strong>seed</strong>: A Python integer to use as random seed.</li>
</ul>
<p><strong>References</strong></p>
<ul>
<li><a href="http://www.jmlr.org/papers/volume15/srivastava14a/srivastava14a.pdf">Dropout: A Simple Way to Prevent Neural Networks from Overfitting</a></li>
</ul>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L462">[source]</a></span></p>
<h3 id="flatten">Flatten</h3>
<pre><code class="python">keras.layers.Flatten(data_format=None)
</code></pre>
<p>Flattens the input. Does not affect the batch size.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>data_format</strong>: A string,
one of <code>channels_last</code> (default) or <code>channels_first</code>.
The ordering of the dimensions in the inputs.
The purpose of this argument is to preserve weight
ordering when switching a model from one data format
to another.
<code>channels_last</code> corresponds to inputs with shape
<code>(batch, ..., channels)</code> while <code>channels_first</code> corresponds to
inputs with shape <code>(batch, channels, ...)</code>.
It defaults to the <code>image_data_format</code> value found in your
Keras config file at <code>~/.keras/keras.json</code>.
If you never set it, then it will be "channels_last".</li>
</ul>
<p><strong>Example</strong></p>
<pre><code class="python">model = Sequential()
model.add(Conv2D(64, (3, 3),
input_shape=(3, 32, 32), padding='same',))
# now: model.output_shape == (None, 64, 32, 32)
model.add(Flatten())
# now: model.output_shape == (None, 65536)
</code></pre>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/engine/input_layer.py#L114">[source]</a></span></p>
<h3 id="input">Input</h3>
<pre><code class="python">keras.engine.input_layer.Input()
</code></pre>
<p><code>Input()</code> is used to instantiate a Keras tensor.</p>
<p>A Keras tensor is a tensor object from the underlying backend
(Theano, TensorFlow or CNTK), which we augment with certain
attributes that allow us to build a Keras model
just by knowing the inputs and outputs of the model.</p>
<p>For instance, if a, b and c are Keras tensors,
it becomes possible to do:
<code>model = Model(input=[a, b], output=c)</code></p>
<p>The added Keras attributes are:
<code>_keras_shape</code>: Integer shape tuple propagated
via Keras-side shape inference.
<code>_keras_history</code>: Last layer applied to the tensor.
the entire layer graph is retrievable from that layer,
recursively.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>shape</strong>: A shape tuple (integer), not including the batch size.
For instance, <code>shape=(32,)</code> indicates that the expected input
will be batches of 32-dimensional vectors.</li>
<li><strong>batch_shape</strong>: A shape tuple (integer), including the batch size.
For instance, <code>batch_shape=(10, 32)</code> indicates that
the expected input will be batches of 10 32-dimensional vectors.
<code>batch_shape=(None, 32)</code> indicates batches of an arbitrary number
of 32-dimensional vectors.</li>
<li><strong>name</strong>: An optional name string for the layer.
Should be unique in a model (do not reuse the same name twice).
It will be autogenerated if it isn't provided.</li>
<li><strong>dtype</strong>: The data type expected by the input, as a string
(<code>float32</code>, <code>float64</code>, <code>int32</code>...)</li>
<li><strong>sparse</strong>: A boolean specifying whether the placeholder
to be created is sparse.</li>
<li><strong>tensor</strong>: Optional existing tensor to wrap into the <code>Input</code> layer.
If set, the layer will not create a placeholder tensor.</li>
</ul>
<p><strong>Returns</strong></p>
<p>A tensor.</p>
<p><strong>Example</strong></p>
<pre><code class="python"># this is a logistic regression in Keras
x = Input(shape=(32,))
y = Dense(16, activation='softmax')(x)
model = Model(x, y)
</code></pre>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L311">[source]</a></span></p>
<h3 id="reshape">Reshape</h3>
<pre><code class="python">keras.layers.Reshape(target_shape)
</code></pre>
<p>Reshapes an output to a certain shape.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>target_shape</strong>: target shape. Tuple of integers.
Does not include the batch axis.</li>
</ul>
<p><strong>Input shape</strong></p>
<p>Arbitrary, although all dimensions in the input shaped must be fixed.
Use the keyword argument <code>input_shape</code>
(tuple of integers, does not include the batch axis)
when using this layer as the first layer in a model.</p>
<p><strong>Output shape</strong></p>
<p><code>(batch_size,) + target_shape</code></p>
<p><strong>Example</strong></p>
<pre><code class="python"># as first layer in a Sequential model
model = Sequential()
model.add(Reshape((3, 4), input_shape=(12,)))
# now: model.output_shape == (None, 3, 4)
# note: `None` is the batch dimension
# as intermediate layer in a Sequential model
model.add(Reshape((6, 2)))
# now: model.output_shape == (None, 6, 2)
# also supports shape inference using `-1` as dimension
model.add(Reshape((-1, 2, 2)))
# now: model.output_shape == (None, 3, 2, 2)
</code></pre>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L410">[source]</a></span></p>
<h3 id="permute">Permute</h3>
<pre><code class="python">keras.layers.Permute(dims)
</code></pre>
<p>Permutes the dimensions of the input according to a given pattern.</p>
<p>Useful for e.g. connecting RNNs and convnets together.</p>
<p><strong>Example</strong></p>
<pre><code class="python">model = Sequential()
model.add(Permute((2, 1), input_shape=(10, 64)))
# now: model.output_shape == (None, 64, 10)
# note: `None` is the batch dimension
</code></pre>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>dims</strong>: Tuple of integers. Permutation pattern, does not include the
samples dimension. Indexing starts at 1.
For instance, <code>(2, 1)</code> permutes the first and second dimension
of the input.</li>
</ul>
<p><strong>Input shape</strong></p>
<p>Arbitrary. Use the keyword argument <code>input_shape</code>
(tuple of integers, does not include the samples axis)
when using this layer as the first layer in a model.</p>
<p><strong>Output shape</strong></p>
<p>Same as the input shape, but with the dimensions re-ordered according
to the specified pattern.</p>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L524">[source]</a></span></p>
<h3 id="repeatvector">RepeatVector</h3>
<pre><code class="python">keras.layers.RepeatVector(n)
</code></pre>
<p>Repeats the input n times.</p>
<p><strong>Example</strong></p>
<pre><code class="python">model = Sequential()
model.add(Dense(32, input_dim=32))
# now: model.output_shape == (None, 32)
# note: `None` is the batch dimension
model.add(RepeatVector(3))
# now: model.output_shape == (None, 3, 32)
</code></pre>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>n</strong>: integer, repetition factor.</li>
</ul>
<p><strong>Input shape</strong></p>
<p>2D tensor of shape <code>(num_samples, features)</code>.</p>
<p><strong>Output shape</strong></p>
<p>3D tensor of shape <code>(num_samples, n, features)</code>.</p>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L566">[source]</a></span></p>
<h3 id="lambda">Lambda</h3>
<pre><code class="python">keras.layers.Lambda(function, output_shape=None, mask=None, arguments=None)
</code></pre>
<p>Wraps arbitrary expression as a <code>Layer</code> object.</p>
<p><strong>Examples</strong></p>
<pre><code class="python"># add a x -> x^2 layer
model.add(Lambda(lambda x: x ** 2))
</code></pre>
<pre><code class="python"># add a layer that returns the concatenation
# of the positive part of the input and
# the opposite of the negative part
def antirectifier(x):
x -= K.mean(x, axis=1, keepdims=True)
x = K.l2_normalize(x, axis=1)
pos = K.relu(x)
neg = K.relu(-x)
return K.concatenate([pos, neg], axis=1)
def antirectifier_output_shape(input_shape):
shape = list(input_shape)
assert len(shape) == 2 # only valid for 2D tensors
shape[-1] *= 2
return tuple(shape)
model.add(Lambda(antirectifier,
output_shape=antirectifier_output_shape))
</code></pre>
<pre><code class="python"># add a layer that returns the hadamard product
# and sum of it from two input tensors
def hadamard_product_sum(tensors):
out1 = tensors[0] * tensors[1]
out2 = K.sum(out1, axis=-1)
return [out1, out2]
def hadamard_product_sum_output_shape(input_shapes):
shape1 = list(input_shapes[0])
shape2 = list(input_shapes[1])
assert shape1 == shape2 # else hadamard product isn't possible
return [tuple(shape1), tuple(shape2[:-1])]
x1 = Dense(32)(input_1)
x2 = Dense(32)(input_2)
layer = Lambda(hadamard_product_sum, hadamard_product_sum_output_shape)
x_hadamard, x_sum = layer([x1, x2])
</code></pre>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>function</strong>: The function to be evaluated.
Takes input tensor or list of tensors as first argument.</li>
<li><strong>output_shape</strong>: Expected output shape from function.
Only relevant when using Theano.
Can be a tuple or function.
If a tuple, it only specifies the first dimension onward;
sample dimension is assumed either the same as the input:
<code>output_shape = (input_shape[0], ) + output_shape</code>
or, the input is <code>None</code> and
the sample dimension is also <code>None</code>:
<code>output_shape = (None, ) + output_shape</code>
If a function, it specifies the entire shape as a function of the
input shape: <code>output_shape = f(input_shape)</code></li>
<li><strong>mask</strong>: Either None (indicating no masking) or a Tensor indicating the
input mask for Embedding.</li>
<li><strong>arguments</strong>: optional dictionary of keyword arguments to be passed
to the function.</li>
</ul>
<p><strong>Input shape</strong></p>
<p>Arbitrary. Use the keyword argument input_shape
(tuple of integers, does not include the samples axis)
when using this layer as the first layer in a model.</p>
<p><strong>Output shape</strong></p>
<p>Specified by <code>output_shape</code> argument
(or auto-inferred when using TensorFlow or CNTK).</p>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L940">[source]</a></span></p>
<h3 id="activityregularization">ActivityRegularization</h3>
<pre><code class="python">keras.layers.ActivityRegularization(l1=0.0, l2=0.0)
</code></pre>
<p>Layer that applies an update to the cost function based input activity.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>l1</strong>: L1 regularization factor (positive float).</li>
<li><strong>l2</strong>: L2 regularization factor (positive float).</li>
</ul>
<p><strong>Input shape</strong></p>
<p>Arbitrary. Use the keyword argument <code>input_shape</code>
(tuple of integers, does not include the samples axis)
when using this layer as the first layer in a model.</p>
<p><strong>Output shape</strong></p>
<p>Same shape as input.</p>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L28">[source]</a></span></p>
<h3 id="masking">Masking</h3>
<pre><code class="python">keras.layers.Masking(mask_value=0.0)
</code></pre>
<p>Masks a sequence by using a mask value to skip timesteps.</p>
<p>If all features for a given sample timestep are equal to <code>mask_value</code>,
then the sample timestep will be masked (skipped) in all downstream layers
(as long as they support masking).</p>
<p>If any downstream layer does not support masking yet receives such
an input mask, an exception will be raised.</p>
<p><strong>Example</strong></p>
<p>Consider a Numpy data array <code>x</code> of shape <code>(samples, timesteps, features)</code>,
to be fed to an LSTM layer.
You want to mask sample #0 at timestep #3, and sample #2 at timestep #5,
because you lack features for these sample timesteps. You can do:</p>
<ul>
<li>set <code>x[0, 3, :] = 0.</code> and <code>x[2, 5, :] = 0.</code></li>
<li>insert a <code>Masking</code> layer with <code>mask_value=0.</code> before the LSTM layer:</li>
</ul>
<pre><code class="python">model = Sequential()
model.add(Masking(mask_value=0., input_shape=(timesteps, features)))
model.add(LSTM(32))
</code></pre>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>mask_value</strong>: Either None or mask value to skip</li>
</ul>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L141">[source]</a></span></p>
<h3 id="spatialdropout1d">SpatialDropout1D</h3>
<pre><code class="python">keras.layers.SpatialDropout1D(rate)
</code></pre>
<p>Spatial 1D version of Dropout.</p>
<p>This version performs the same function as Dropout, however it drops
entire 1D feature maps instead of individual elements. If adjacent frames
within feature maps are strongly correlated (as is normally the case in
early convolution layers) then regular dropout will not regularize the
activations and will otherwise just result in an effective learning rate
decrease. In this case, SpatialDropout1D will help promote independence
between feature maps and should be used instead.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>rate</strong>: float between 0 and 1. Fraction of the input units to drop.</li>
</ul>
<p><strong>Input shape</strong></p>
<p>3D tensor with shape:
<code>(samples, timesteps, channels)</code></p>
<p><strong>Output shape</strong></p>
<p>Same as input</p>
<p><strong>References</strong></p>
<ul>
<li><a href="https://arxiv.org/abs/1411.4280">Efficient Object Localization Using Convolutional Networks</a></li>
</ul>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L178">[source]</a></span></p>
<h3 id="spatialdropout2d">SpatialDropout2D</h3>
<pre><code class="python">keras.layers.SpatialDropout2D(rate, data_format=None)
</code></pre>
<p>Spatial 2D version of Dropout.</p>
<p>This version performs the same function as Dropout, however it drops
entire 2D feature maps instead of individual elements. If adjacent pixels
within feature maps are strongly correlated (as is normally the case in
early convolution layers) then regular dropout will not regularize the
activations and will otherwise just result in an effective learning rate
decrease. In this case, SpatialDropout2D will help promote independence
between feature maps and should be used instead.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>rate</strong>: float between 0 and 1. Fraction of the input units to drop.</li>
<li><strong>data_format</strong>: 'channels_first' or 'channels_last'.
In 'channels_first' mode, the channels dimension
(the depth) is at index 1,
in 'channels_last' mode is it at index 3.
It defaults to the <code>image_data_format</code> value found in your
Keras config file at <code>~/.keras/keras.json</code>.
If you never set it, then it will be "channels_last".</li>
</ul>
<p><strong>Input shape</strong></p>
<p>4D tensor with shape:
<code>(samples, channels, rows, cols)</code> if data_format='channels_first'
or 4D tensor with shape:
<code>(samples, rows, cols, channels)</code> if data_format='channels_last'.</p>
<p><strong>Output shape</strong></p>
<p>Same as input</p>
<p><strong>References</strong></p>
<ul>
<li><a href="https://arxiv.org/abs/1411.4280">Efficient Object Localization Using Convolutional Networks</a></li>
</ul>
<hr />
<p><span style="float:right;"><a href="https://github.com/keras-team/keras/blob/master/keras/layers/core.py#L228">[source]</a></span></p>
<h3 id="spatialdropout3d">SpatialDropout3D</h3>
<pre><code class="python">keras.layers.SpatialDropout3D(rate, data_format=None)
</code></pre>
<p>Spatial 3D version of Dropout.</p>
<p>This version performs the same function as Dropout, however it drops
entire 3D feature maps instead of individual elements. If adjacent voxels
within feature maps are strongly correlated (as is normally the case in
early convolution layers) then regular dropout will not regularize the
activations and will otherwise just result in an effective learning rate
decrease. In this case, SpatialDropout3D will help promote independence
between feature maps and should be used instead.</p>
<p><strong>Arguments</strong></p>
<ul>
<li><strong>rate</strong>: float between 0 and 1. Fraction of the input units to drop.</li>
<li><strong>data_format</strong>: 'channels_first' or 'channels_last'.
In 'channels_first' mode, the channels dimension (the depth)
is at index 1, in 'channels_last' mode is it at index 4.
It defaults to the <code>image_data_format</code> value found in your
Keras config file at <code>~/.keras/keras.json</code>.
If you never set it, then it will be "channels_last".</li>
</ul>
<p><strong>Input shape</strong></p>
<p>5D tensor with shape:
<code>(samples, channels, dim1, dim2, dim3)</code> if data_format='channels_first'
or 5D tensor with shape:
<code>(samples, dim1, dim2, dim3, channels)</code> if data_format='channels_last'.</p>
<p><strong>Output shape</strong></p>
<p>Same as input</p>
<p><strong>References</strong></p>
<ul>
<li><a href="https://arxiv.org/abs/1411.4280">Efficient Object Localization Using Convolutional Networks</a></li>
</ul>
</div>
</div>
<footer>
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
<a href="../convolutional/" class="btn btn-neutral float-right" title="Convolutional Layers">Next <span class="icon icon-circle-arrow-right"></span></a>
<a href="../about-keras-layers/" class="btn btn-neutral" title="About Keras layers"><span class="icon icon-circle-arrow-left"></span> Previous</a>
</div>
<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><a href="../about-keras-layers/" style="color: #fcfcfc;">« Previous</a></span>
<span style="margin-left: 15px"><a href="../convolutional/" style="color: #fcfcfc">Next »</a></span>
</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>
|
