-
Notifications
You must be signed in to change notification settings - Fork 81
/
Copy pathneural_networks_supervised.html
1118 lines (888 loc) · 66.5 KB
/
neural_networks_supervised.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
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
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html>
<html lang="en" data-content_root="../" >
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" /><meta name="viewport" content="width=device-width, initial-scale=1" />
<meta property="og:title" content="1.17. Neural network models (supervised)" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/modules/neural_networks_supervised.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="Multi-layer Perceptron: Multi-layer Perceptron (MLP) is a supervised learning algorithm that learns a function f: R^m \rightarrow R^o by training on a dataset, where m is the number of dimensions f..." />
<meta property="og:image" content="https://scikit-learn/stable/_images/multilayerperceptron_network.png" />
<meta property="og:image:alt" content="scikit-learn" />
<meta name="description" content="Multi-layer Perceptron: Multi-layer Perceptron (MLP) is a supervised learning algorithm that learns a function f: R^m \rightarrow R^o by training on a dataset, where m is the number of dimensions f..." />
<title>1.17. Neural network models (supervised) — scikit-learn 1.6.1 documentation</title>
<script data-cfasync="false">
document.documentElement.dataset.mode = localStorage.getItem("mode") || "";
document.documentElement.dataset.theme = localStorage.getItem("theme") || "";
</script>
<!--
this give us a css class that will be invisible only if js is disabled
-->
<noscript>
<style>
.pst-js-only { display: none !important; }
</style>
</noscript>
<!-- Loaded before other Sphinx assets -->
<link href="../_static/styles/theme.css?digest=8878045cc6db502f8baf" rel="stylesheet" />
<link href="../_static/styles/pydata-sphinx-theme.css?digest=8878045cc6db502f8baf" rel="stylesheet" />
<link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=a746c00c" />
<link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
<link rel="stylesheet" type="text/css" href="../_static/plot_directive.css" />
<link rel="stylesheet" type="text/css" href="https://fonts.googleapis.com/css?family=Vibur" />
<link rel="stylesheet" type="text/css" href="../_static/jupyterlite_sphinx.css?v=e3ca86de" />
<link rel="stylesheet" type="text/css" href="../_static/sg_gallery.css?v=d2d258e8" />
<link rel="stylesheet" type="text/css" href="../_static/sg_gallery-binder.css?v=f4aeca0c" />
<link rel="stylesheet" type="text/css" href="../_static/sg_gallery-dataframe.css?v=2082cf3c" />
<link rel="stylesheet" type="text/css" href="../_static/sg_gallery-rendered-html.css?v=1277b6f3" />
<link rel="stylesheet" type="text/css" href="../_static/sphinx-design.min.css?v=95c83b7e" />
<link rel="stylesheet" type="text/css" href="../_static/styles/colors.css?v=cc94ab7d" />
<link rel="stylesheet" type="text/css" href="../_static/styles/custom.css?v=d67e4bb0" />
<!-- So that users can add custom icons -->
<script src="../_static/scripts/fontawesome.js?digest=8878045cc6db502f8baf"></script>
<!-- Pre-loaded scripts that we'll load fully later -->
<link rel="preload" as="script" href="../_static/scripts/bootstrap.js?digest=8878045cc6db502f8baf" />
<link rel="preload" as="script" href="../_static/scripts/pydata-sphinx-theme.js?digest=8878045cc6db502f8baf" />
<script src="../_static/documentation_options.js?v=d6a008b6"></script>
<script src="../_static/doctools.js?v=9a2dae69"></script>
<script src="../_static/sphinx_highlight.js?v=dc90522c"></script>
<script src="../_static/clipboard.min.js?v=a7894cd8"></script>
<script src="../_static/copybutton.js?v=97f0b27d"></script>
<script src="../_static/jupyterlite_sphinx.js?v=d6bdf5f8"></script>
<script src="../_static/design-tabs.js?v=f930bc37"></script>
<script data-domain="scikit-learn.org" defer="defer" src="https://views.scientific-python.org/js/script.js"></script>
<script async="async" src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-chtml.js"></script>
<script>DOCUMENTATION_OPTIONS.pagename = 'modules/neural_networks_supervised';</script>
<script>
DOCUMENTATION_OPTIONS.theme_version = '0.16.1';
DOCUMENTATION_OPTIONS.theme_switcher_json_url = 'https://scikit-learn.org/dev/_static/versions.json';
DOCUMENTATION_OPTIONS.theme_switcher_version_match = '1.6.1';
DOCUMENTATION_OPTIONS.show_version_warning_banner =
true;
</script>
<script src="../_static/scripts/dropdown.js?v=e2048168"></script>
<script src="../_static/scripts/version-switcher.js?v=a6dd8357"></script>
<script src="../_static/scripts/sg_plotly_resize.js?v=eeb41cab"></script>
<link rel="canonical" href="https://scikit-learn.org/stable/modules/neural_networks_supervised.html" />
<link rel="icon" href="../_static/favicon.ico"/>
<link rel="author" title="About these documents" href="../about.html" />
<link rel="search" title="Search" href="../search.html" />
<link rel="next" title="2. Unsupervised learning" href="../unsupervised_learning.html" />
<link rel="prev" title="1.16. Probability calibration" href="calibration.html" />
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<meta name="docsearch:language" content="en"/>
<meta name="docsearch:version" content="1.6" />
</head>
<body data-bs-spy="scroll" data-bs-target=".bd-toc-nav" data-offset="180" data-bs-root-margin="0px 0px -60%" data-default-mode="">
<div id="pst-skip-link" class="skip-link d-print-none"><a href="#main-content">Skip to main content</a></div>
<div id="pst-scroll-pixel-helper"></div>
<button type="button" class="btn rounded-pill" id="pst-back-to-top">
<i class="fa-solid fa-arrow-up"></i>Back to top</button>
<dialog id="pst-search-dialog">
<form class="bd-search d-flex align-items-center"
action="../search.html"
method="get">
<i class="fa-solid fa-magnifying-glass"></i>
<input type="search"
class="form-control"
name="q"
placeholder="Search the docs ..."
aria-label="Search the docs ..."
autocomplete="off"
autocorrect="off"
autocapitalize="off"
spellcheck="false"/>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd>K</kbd></span>
</form>
</dialog>
<div class="pst-async-banner-revealer d-none">
<aside id="bd-header-version-warning" class="d-none d-print-none" aria-label="Version warning"></aside>
</div>
<header class="bd-header navbar navbar-expand-lg bd-navbar d-print-none">
<div class="bd-header__inner bd-page-width">
<button class="pst-navbar-icon sidebar-toggle primary-toggle" aria-label="Site navigation">
<span class="fa-solid fa-bars"></span>
</button>
<div class=" navbar-header-items__start">
<div class="navbar-item">
<a class="navbar-brand logo" href="../index.html">
<img src="../_static/scikit-learn-logo-small.png" class="logo__image only-light" alt="scikit-learn homepage"/>
<img src="../_static/scikit-learn-logo-small.png" class="logo__image only-dark pst-js-only" alt="scikit-learn homepage"/>
</a></div>
</div>
<div class=" navbar-header-items">
<div class="me-auto navbar-header-items__center">
<div class="navbar-item">
<nav>
<ul class="bd-navbar-elements navbar-nav">
<li class="nav-item ">
<a class="nav-link nav-internal" href="../install.html">
Install
</a>
</li>
<li class="nav-item current active">
<a class="nav-link nav-internal" href="../user_guide.html">
User Guide
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../api/index.html">
API
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../auto_examples/index.html">
Examples
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://blog.scikit-learn.org/">
Community
</a>
</li>
<li class="nav-item dropdown">
<button class="btn dropdown-toggle nav-item" type="button"
data-bs-toggle="dropdown" aria-expanded="false"
aria-controls="pst-nav-more-links">
More
</button>
<ul id="pst-nav-more-links" class="dropdown-menu">
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../getting_started.html">
Getting Started
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../whats_new.html">
Release History
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../glossary.html">
Glossary
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-external" href="https://scikit-learn.org/dev/developers/index.html">
Development
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../faq.html">
FAQ
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../support.html">
Support
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../related_projects.html">
Related Projects
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../roadmap.html">
Roadmap
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../governance.html">
Governance
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="../about.html">
About us
</a>
</li>
</ul>
</li>
</ul>
</nav></div>
</div>
<div class="navbar-header-items__end">
<div class="navbar-item navbar-persistent--container">
<button class="btn btn-sm pst-navbar-icon search-button search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass fa-lg"></i>
</button>
</div>
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/scikit-learn/scikit-learn" title="GitHub" class="nav-link pst-navbar-icon" rel="noopener" target="_blank" data-bs-toggle="tooltip" data-bs-placement="bottom"><i class="fa-brands fa-square-github fa-lg" aria-hidden="true"></i>
<span class="sr-only">GitHub</span></a>
</li>
</ul></div>
<div class="navbar-item">
<div class="version-switcher__container dropdown pst-js-only">
<button id="pst-version-switcher-button-2"
type="button"
class="version-switcher__button btn btn-sm dropdown-toggle"
data-bs-toggle="dropdown"
aria-haspopup="listbox"
aria-controls="pst-version-switcher-list-2"
aria-label="Version switcher list"
>
Choose version <!-- this text may get changed later by javascript -->
<span class="caret"></span>
</button>
<div id="pst-version-switcher-list-2"
class="version-switcher__menu dropdown-menu list-group-flush py-0"
role="listbox" aria-labelledby="pst-version-switcher-button-2">
<!-- dropdown will be populated by javascript on page load -->
</div>
</div></div>
</div>
</div>
<div class="navbar-persistent--mobile">
<button class="btn btn-sm pst-navbar-icon search-button search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass fa-lg"></i>
</button>
</div>
<button class="pst-navbar-icon sidebar-toggle secondary-toggle" aria-label="On this page">
<span class="fa-solid fa-outdent"></span>
</button>
</div>
</header>
<div class="bd-container">
<div class="bd-container__inner bd-page-width">
<dialog id="pst-primary-sidebar-modal"></dialog>
<div id="pst-primary-sidebar" class="bd-sidebar-primary bd-sidebar">
<div class="sidebar-header-items sidebar-primary__section">
<div class="sidebar-header-items__center">
<div class="navbar-item">
<nav>
<ul class="bd-navbar-elements navbar-nav">
<li class="nav-item ">
<a class="nav-link nav-internal" href="../install.html">
Install
</a>
</li>
<li class="nav-item current active">
<a class="nav-link nav-internal" href="../user_guide.html">
User Guide
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../api/index.html">
API
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../auto_examples/index.html">
Examples
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://blog.scikit-learn.org/">
Community
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../getting_started.html">
Getting Started
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../whats_new.html">
Release History
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../glossary.html">
Glossary
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://scikit-learn.org/dev/developers/index.html">
Development
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../faq.html">
FAQ
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../support.html">
Support
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../related_projects.html">
Related Projects
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../roadmap.html">
Roadmap
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../governance.html">
Governance
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="../about.html">
About us
</a>
</li>
</ul>
</nav></div>
</div>
<div class="sidebar-header-items__end">
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/scikit-learn/scikit-learn" title="GitHub" class="nav-link pst-navbar-icon" rel="noopener" target="_blank" data-bs-toggle="tooltip" data-bs-placement="bottom"><i class="fa-brands fa-square-github fa-lg" aria-hidden="true"></i>
<span class="sr-only">GitHub</span></a>
</li>
</ul></div>
<div class="navbar-item">
<div class="version-switcher__container dropdown pst-js-only">
<button id="pst-version-switcher-button-3"
type="button"
class="version-switcher__button btn btn-sm dropdown-toggle"
data-bs-toggle="dropdown"
aria-haspopup="listbox"
aria-controls="pst-version-switcher-list-3"
aria-label="Version switcher list"
>
Choose version <!-- this text may get changed later by javascript -->
<span class="caret"></span>
</button>
<div id="pst-version-switcher-list-3"
class="version-switcher__menu dropdown-menu list-group-flush py-0"
role="listbox" aria-labelledby="pst-version-switcher-button-3">
<!-- dropdown will be populated by javascript on page load -->
</div>
</div></div>
</div>
</div>
<div class="sidebar-primary-items__start sidebar-primary__section">
<div class="sidebar-primary-item">
<nav class="bd-docs-nav bd-links"
aria-label="Section Navigation">
<p class="bd-links__title" role="heading" aria-level="1">Section Navigation</p>
<div class="bd-toc-item navbar-nav"><ul class="current nav bd-sidenav">
<li class="toctree-l1 current active has-children"><a class="reference internal" href="../supervised_learning.html">1. Supervised learning</a><details open="open"><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul class="current">
<li class="toctree-l2"><a class="reference internal" href="linear_model.html">1.1. Linear Models</a></li>
<li class="toctree-l2"><a class="reference internal" href="lda_qda.html">1.2. Linear and Quadratic Discriminant Analysis</a></li>
<li class="toctree-l2"><a class="reference internal" href="kernel_ridge.html">1.3. Kernel ridge regression</a></li>
<li class="toctree-l2"><a class="reference internal" href="svm.html">1.4. Support Vector Machines</a></li>
<li class="toctree-l2"><a class="reference internal" href="sgd.html">1.5. Stochastic Gradient Descent</a></li>
<li class="toctree-l2"><a class="reference internal" href="neighbors.html">1.6. Nearest Neighbors</a></li>
<li class="toctree-l2"><a class="reference internal" href="gaussian_process.html">1.7. Gaussian Processes</a></li>
<li class="toctree-l2"><a class="reference internal" href="cross_decomposition.html">1.8. Cross decomposition</a></li>
<li class="toctree-l2"><a class="reference internal" href="naive_bayes.html">1.9. Naive Bayes</a></li>
<li class="toctree-l2"><a class="reference internal" href="tree.html">1.10. Decision Trees</a></li>
<li class="toctree-l2"><a class="reference internal" href="ensemble.html">1.11. Ensembles: Gradient boosting, random forests, bagging, voting, stacking</a></li>
<li class="toctree-l2"><a class="reference internal" href="multiclass.html">1.12. Multiclass and multioutput algorithms</a></li>
<li class="toctree-l2"><a class="reference internal" href="feature_selection.html">1.13. Feature selection</a></li>
<li class="toctree-l2"><a class="reference internal" href="semi_supervised.html">1.14. Semi-supervised learning</a></li>
<li class="toctree-l2"><a class="reference internal" href="isotonic.html">1.15. Isotonic regression</a></li>
<li class="toctree-l2"><a class="reference internal" href="calibration.html">1.16. Probability calibration</a></li>
<li class="toctree-l2 current active"><a class="current reference internal" href="#">1.17. Neural network models (supervised)</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../unsupervised_learning.html">2. Unsupervised learning</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="mixture.html">2.1. Gaussian mixture models</a></li>
<li class="toctree-l2"><a class="reference internal" href="manifold.html">2.2. Manifold learning</a></li>
<li class="toctree-l2"><a class="reference internal" href="clustering.html">2.3. Clustering</a></li>
<li class="toctree-l2"><a class="reference internal" href="biclustering.html">2.4. Biclustering</a></li>
<li class="toctree-l2"><a class="reference internal" href="decomposition.html">2.5. Decomposing signals in components (matrix factorization problems)</a></li>
<li class="toctree-l2"><a class="reference internal" href="covariance.html">2.6. Covariance estimation</a></li>
<li class="toctree-l2"><a class="reference internal" href="outlier_detection.html">2.7. Novelty and Outlier Detection</a></li>
<li class="toctree-l2"><a class="reference internal" href="density.html">2.8. Density Estimation</a></li>
<li class="toctree-l2"><a class="reference internal" href="neural_networks_unsupervised.html">2.9. Neural network models (unsupervised)</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../model_selection.html">3. Model selection and evaluation</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="cross_validation.html">3.1. Cross-validation: evaluating estimator performance</a></li>
<li class="toctree-l2"><a class="reference internal" href="grid_search.html">3.2. Tuning the hyper-parameters of an estimator</a></li>
<li class="toctree-l2"><a class="reference internal" href="classification_threshold.html">3.3. Tuning the decision threshold for class prediction</a></li>
<li class="toctree-l2"><a class="reference internal" href="model_evaluation.html">3.4. Metrics and scoring: quantifying the quality of predictions</a></li>
<li class="toctree-l2"><a class="reference internal" href="learning_curve.html">3.5. Validation curves: plotting scores to evaluate models</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../inspection.html">4. Inspection</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="partial_dependence.html">4.1. Partial Dependence and Individual Conditional Expectation plots</a></li>
<li class="toctree-l2"><a class="reference internal" href="permutation_importance.html">4.2. Permutation feature importance</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="../visualizations.html">5. Visualizations</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../data_transforms.html">6. Dataset transformations</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="compose.html">6.1. Pipelines and composite estimators</a></li>
<li class="toctree-l2"><a class="reference internal" href="feature_extraction.html">6.2. Feature extraction</a></li>
<li class="toctree-l2"><a class="reference internal" href="preprocessing.html">6.3. Preprocessing data</a></li>
<li class="toctree-l2"><a class="reference internal" href="impute.html">6.4. Imputation of missing values</a></li>
<li class="toctree-l2"><a class="reference internal" href="unsupervised_reduction.html">6.5. Unsupervised dimensionality reduction</a></li>
<li class="toctree-l2"><a class="reference internal" href="random_projection.html">6.6. Random Projection</a></li>
<li class="toctree-l2"><a class="reference internal" href="kernel_approximation.html">6.7. Kernel Approximation</a></li>
<li class="toctree-l2"><a class="reference internal" href="metrics.html">6.8. Pairwise metrics, Affinities and Kernels</a></li>
<li class="toctree-l2"><a class="reference internal" href="preprocessing_targets.html">6.9. Transforming the prediction target (<code class="docutils literal notranslate"><span class="pre">y</span></code>)</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../datasets.html">7. Dataset loading utilities</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="../datasets/toy_dataset.html">7.1. Toy datasets</a></li>
<li class="toctree-l2"><a class="reference internal" href="../datasets/real_world.html">7.2. Real world datasets</a></li>
<li class="toctree-l2"><a class="reference internal" href="../datasets/sample_generators.html">7.3. Generated datasets</a></li>
<li class="toctree-l2"><a class="reference internal" href="../datasets/loading_other_datasets.html">7.4. Loading other datasets</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../computing.html">8. Computing with scikit-learn</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="../computing/scaling_strategies.html">8.1. Strategies to scale computationally: bigger data</a></li>
<li class="toctree-l2"><a class="reference internal" href="../computing/computational_performance.html">8.2. Computational Performance</a></li>
<li class="toctree-l2"><a class="reference internal" href="../computing/parallelism.html">8.3. Parallelism, resource management, and configuration</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="../model_persistence.html">9. Model persistence</a></li>
<li class="toctree-l1"><a class="reference internal" href="../common_pitfalls.html">10. Common pitfalls and recommended practices</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../dispatching.html">11. Dispatching</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="array_api.html">11.1. Array API support (experimental)</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="../machine_learning_map.html">12. Choosing the right estimator</a></li>
<li class="toctree-l1"><a class="reference internal" href="../presentations.html">13. External Resources, Videos and Talks</a></li>
</ul>
</div>
</nav></div>
</div>
<div class="sidebar-primary-items__end sidebar-primary__section">
</div>
</div>
<main id="main-content" class="bd-main" role="main">
<div class="bd-content">
<div class="bd-article-container">
<div class="bd-header-article d-print-none">
<div class="header-article-items header-article__inner">
<div class="header-article-items__start">
<div class="header-article-item">
<nav aria-label="Breadcrumb" class="d-print-none">
<ul class="bd-breadcrumbs">
<li class="breadcrumb-item breadcrumb-home">
<a href="../index.html" class="nav-link" aria-label="Home">
<i class="fa-solid fa-home"></i>
</a>
</li>
<li class="breadcrumb-item"><a href="../user_guide.html" class="nav-link">User Guide</a></li>
<li class="breadcrumb-item"><a href="../supervised_learning.html" class="nav-link"><span class="section-number">1. </span>Supervised learning</a></li>
<li class="breadcrumb-item active" aria-current="page"><span class="ellipsis"><span class="section-number">1.17. </span>Neural network models (supervised)</span></li>
</ul>
</nav>
</div>
</div>
</div>
</div>
<div id="searchbox"></div>
<article class="bd-article">
<section id="neural-network-models-supervised">
<span id="neural-networks-supervised"></span><h1><span class="section-number">1.17. </span>Neural network models (supervised)<a class="headerlink" href="#neural-network-models-supervised" title="Link to this heading">#</a></h1>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>This implementation is not intended for large-scale applications. In particular,
scikit-learn offers no GPU support. For much faster, GPU-based implementations,
as well as frameworks offering much more flexibility to build deep learning
architectures, see <a class="reference internal" href="../related_projects.html#related-projects"><span class="std std-ref">Related Projects</span></a>.</p>
</div>
<section id="multi-layer-perceptron">
<span id="multilayer-perceptron"></span><h2><span class="section-number">1.17.1. </span>Multi-layer Perceptron<a class="headerlink" href="#multi-layer-perceptron" title="Link to this heading">#</a></h2>
<p><strong>Multi-layer Perceptron (MLP)</strong> is a supervised learning algorithm that learns
a function <span class="math notranslate nohighlight">\(f: R^m \rightarrow R^o\)</span> by training on a dataset,
where <span class="math notranslate nohighlight">\(m\)</span> is the number of dimensions for input and <span class="math notranslate nohighlight">\(o\)</span> is the
number of dimensions for output. Given a set of features <span class="math notranslate nohighlight">\(X = {x_1, x_2, ..., x_m}\)</span>
and a target <span class="math notranslate nohighlight">\(y\)</span>, it can learn a non-linear function approximator for either
classification or regression. It is different from logistic regression, in that
between the input and the output layer, there can be one or more non-linear
layers, called hidden layers. Figure 1 shows a one hidden layer MLP with scalar
output.</p>
<figure class="align-center" id="id3">
<a class="reference internal image-reference" href="../_images/multilayerperceptron_network.png"><img alt="../_images/multilayerperceptron_network.png" src="../_images/multilayerperceptron_network.png" style="width: 469.79999999999995px; height: 510.59999999999997px;" />
</a>
<figcaption>
<p><span class="caption-text"><strong>Figure 1 : One hidden layer MLP.</strong></span><a class="headerlink" href="#id3" title="Link to this image">#</a></p>
</figcaption>
</figure>
<p>The leftmost layer, known as the input layer, consists of a set of neurons
<span class="math notranslate nohighlight">\(\{x_i | x_1, x_2, ..., x_m\}\)</span> representing the input features. Each
neuron in the hidden layer transforms the values from the previous layer with
a weighted linear summation <span class="math notranslate nohighlight">\(w_1x_1 + w_2x_2 + ... + w_mx_m\)</span>, followed
by a non-linear activation function <span class="math notranslate nohighlight">\(g(\cdot):R \rightarrow R\)</span> - like
the hyperbolic tan function. The output layer receives the values from the
last hidden layer and transforms them into output values.</p>
<p>The module contains the public attributes <code class="docutils literal notranslate"><span class="pre">coefs_</span></code> and <code class="docutils literal notranslate"><span class="pre">intercepts_</span></code>.
<code class="docutils literal notranslate"><span class="pre">coefs_</span></code> is a list of weight matrices, where weight matrix at index
<span class="math notranslate nohighlight">\(i\)</span> represents the weights between layer <span class="math notranslate nohighlight">\(i\)</span> and layer
<span class="math notranslate nohighlight">\(i+1\)</span>. <code class="docutils literal notranslate"><span class="pre">intercepts_</span></code> is a list of bias vectors, where the vector
at index <span class="math notranslate nohighlight">\(i\)</span> represents the bias values added to layer <span class="math notranslate nohighlight">\(i+1\)</span>.</p>
<details class="sd-sphinx-override sd-dropdown sd-card sd-mb-3" id="advantages-and-disadvantages-of-multi-layer-perceptron">
<summary class="sd-summary-title sd-card-header">
<span class="sd-summary-text">Advantages and disadvantages of Multi-layer Perceptron<a class="headerlink" href="#advantages-and-disadvantages-of-multi-layer-perceptron" title="Link to this dropdown">#</a></span><span class="sd-summary-state-marker sd-summary-chevron-right"><svg version="1.1" width="1.5em" height="1.5em" class="sd-octicon sd-octicon-chevron-right" viewBox="0 0 24 24" aria-hidden="true"><path d="M8.72 18.78a.75.75 0 0 1 0-1.06L14.44 12 8.72 6.28a.751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018l6.25 6.25a.75.75 0 0 1 0 1.06l-6.25 6.25a.75.75 0 0 1-1.06 0Z"></path></svg></span></summary><div class="sd-summary-content sd-card-body docutils">
<p class="sd-card-text">The advantages of Multi-layer Perceptron are:</p>
<ul class="simple">
<li><p class="sd-card-text">Capability to learn non-linear models.</p></li>
<li><p class="sd-card-text">Capability to learn models in real-time (on-line learning)
using <code class="docutils literal notranslate"><span class="pre">partial_fit</span></code>.</p></li>
</ul>
<p class="sd-card-text">The disadvantages of Multi-layer Perceptron (MLP) include:</p>
<ul class="simple">
<li><p class="sd-card-text">MLP with hidden layers have a non-convex loss function where there exists
more than one local minimum. Therefore different random weight
initializations can lead to different validation accuracy.</p></li>
<li><p class="sd-card-text">MLP requires tuning a number of hyperparameters such as the number of
hidden neurons, layers, and iterations.</p></li>
<li><p class="sd-card-text">MLP is sensitive to feature scaling.</p></li>
</ul>
<p class="sd-card-text">Please see <a class="reference internal" href="#mlp-tips"><span class="std std-ref">Tips on Practical Use</span></a> section that addresses
some of these disadvantages.</p>
</div>
</details></section>
<section id="classification">
<h2><span class="section-number">1.17.2. </span>Classification<a class="headerlink" href="#classification" title="Link to this heading">#</a></h2>
<p>Class <a class="reference internal" href="generated/sklearn.neural_network.MLPClassifier.html#sklearn.neural_network.MLPClassifier" title="sklearn.neural_network.MLPClassifier"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLPClassifier</span></code></a> implements a multi-layer perceptron (MLP) algorithm
that trains using <a class="reference external" href="http://ufldl.stanford.edu/wiki/index.php/Backpropagation_Algorithm">Backpropagation</a>.</p>
<p>MLP trains on two arrays: array X of size (n_samples, n_features), which holds
the training samples represented as floating point feature vectors; and array
y of size (n_samples,), which holds the target values (class labels) for the
training samples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.neural_network</span> <span class="kn">import</span> <span class="n">MLPClassifier</span>
<span class="gp">>>> </span><span class="n">X</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span>
<span class="gp">>>> </span><span class="n">y</span> <span class="o">=</span> <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">1</span><span class="p">]</span>
<span class="gp">>>> </span><span class="n">clf</span> <span class="o">=</span> <span class="n">MLPClassifier</span><span class="p">(</span><span class="n">solver</span><span class="o">=</span><span class="s1">'lbfgs'</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">1e-5</span><span class="p">,</span>
<span class="gp">... </span> <span class="n">hidden_layer_sizes</span><span class="o">=</span><span class="p">(</span><span class="mi">5</span><span class="p">,</span> <span class="mi">2</span><span class="p">),</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span>
<span class="gp">...</span>
<span class="gp">>>> </span><span class="n">clf</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span>
<span class="go">MLPClassifier(alpha=1e-05, hidden_layer_sizes=(5, 2), random_state=1,</span>
<span class="go"> solver='lbfgs')</span>
</pre></div>
</div>
<p>After fitting (training), the model can predict labels for new samples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">clf</span><span class="o">.</span><span class="n">predict</span><span class="p">([[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> <span class="p">[</span><span class="o">-</span><span class="mf">1.</span><span class="p">,</span> <span class="o">-</span><span class="mf">2.</span><span class="p">]])</span>
<span class="go">array([1, 0])</span>
</pre></div>
</div>
<p>MLP can fit a non-linear model to the training data. <code class="docutils literal notranslate"><span class="pre">clf.coefs_</span></code>
contains the weight matrices that constitute the model parameters:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="p">[</span><span class="n">coef</span><span class="o">.</span><span class="n">shape</span> <span class="k">for</span> <span class="n">coef</span> <span class="ow">in</span> <span class="n">clf</span><span class="o">.</span><span class="n">coefs_</span><span class="p">]</span>
<span class="go">[(2, 5), (5, 2), (2, 1)]</span>
</pre></div>
</div>
<p>Currently, <a class="reference internal" href="generated/sklearn.neural_network.MLPClassifier.html#sklearn.neural_network.MLPClassifier" title="sklearn.neural_network.MLPClassifier"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLPClassifier</span></code></a> supports only the
Cross-Entropy loss function, which allows probability estimates by running the
<code class="docutils literal notranslate"><span class="pre">predict_proba</span></code> method.</p>
<p>MLP trains using Backpropagation. More precisely, it trains using some form of
gradient descent and the gradients are calculated using Backpropagation. For
classification, it minimizes the Cross-Entropy loss function, giving a vector
of probability estimates <span class="math notranslate nohighlight">\(P(y|x)\)</span> per sample <span class="math notranslate nohighlight">\(x\)</span>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">clf</span><span class="o">.</span><span class="n">predict_proba</span><span class="p">([[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]])</span>
<span class="go">array([[1.967...e-04, 9.998...-01],</span>
<span class="go"> [1.967...e-04, 9.998...-01]])</span>
</pre></div>
</div>
<p><a class="reference internal" href="generated/sklearn.neural_network.MLPClassifier.html#sklearn.neural_network.MLPClassifier" title="sklearn.neural_network.MLPClassifier"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLPClassifier</span></code></a> supports multi-class classification by
applying <a class="reference external" href="https://en.wikipedia.org/wiki/Softmax_activation_function">Softmax</a>
as the output function.</p>
<p>Further, the model supports <a class="reference internal" href="multiclass.html#multiclass"><span class="std std-ref">multi-label classification</span></a>
in which a sample can belong to more than one class. For each class, the raw
output passes through the logistic function. Values larger or equal to <code class="docutils literal notranslate"><span class="pre">0.5</span></code>
are rounded to <code class="docutils literal notranslate"><span class="pre">1</span></code>, otherwise to <code class="docutils literal notranslate"><span class="pre">0</span></code>. For a predicted output of a sample, the
indices where the value is <code class="docutils literal notranslate"><span class="pre">1</span></code> represents the assigned classes of that sample:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">X</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span>
<span class="gp">>>> </span><span class="n">y</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">]]</span>
<span class="gp">>>> </span><span class="n">clf</span> <span class="o">=</span> <span class="n">MLPClassifier</span><span class="p">(</span><span class="n">solver</span><span class="o">=</span><span class="s1">'lbfgs'</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">1e-5</span><span class="p">,</span>
<span class="gp">... </span> <span class="n">hidden_layer_sizes</span><span class="o">=</span><span class="p">(</span><span class="mi">15</span><span class="p">,),</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span>
<span class="gp">...</span>
<span class="gp">>>> </span><span class="n">clf</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span>
<span class="go">MLPClassifier(alpha=1e-05, hidden_layer_sizes=(15,), random_state=1,</span>
<span class="go"> solver='lbfgs')</span>
<span class="gp">>>> </span><span class="n">clf</span><span class="o">.</span><span class="n">predict</span><span class="p">([[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]])</span>
<span class="go">array([[1, 1]])</span>
<span class="gp">>>> </span><span class="n">clf</span><span class="o">.</span><span class="n">predict</span><span class="p">([[</span><span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]])</span>
<span class="go">array([[0, 1]])</span>
</pre></div>
</div>
<p>See the examples below and the docstring of
<a class="reference internal" href="generated/sklearn.neural_network.MLPClassifier.html#sklearn.neural_network.MLPClassifier.fit" title="sklearn.neural_network.MLPClassifier.fit"><code class="xref py py-meth docutils literal notranslate"><span class="pre">MLPClassifier.fit</span></code></a> for further information.</p>
<p class="rubric">Examples</p>
<ul class="simple">
<li><p><a class="reference internal" href="../auto_examples/neural_networks/plot_mlp_training_curves.html#sphx-glr-auto-examples-neural-networks-plot-mlp-training-curves-py"><span class="std std-ref">Compare Stochastic learning strategies for MLPClassifier</span></a></p></li>
<li><p>See <a class="reference internal" href="../auto_examples/neural_networks/plot_mnist_filters.html#sphx-glr-auto-examples-neural-networks-plot-mnist-filters-py"><span class="std std-ref">Visualization of MLP weights on MNIST</span></a> for
visualized representation of trained weights.</p></li>
</ul>
</section>
<section id="regression">
<h2><span class="section-number">1.17.3. </span>Regression<a class="headerlink" href="#regression" title="Link to this heading">#</a></h2>
<p>Class <a class="reference internal" href="generated/sklearn.neural_network.MLPRegressor.html#sklearn.neural_network.MLPRegressor" title="sklearn.neural_network.MLPRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLPRegressor</span></code></a> implements a multi-layer perceptron (MLP) that
trains using backpropagation with no activation function in the output layer,
which can also be seen as using the identity function as activation function.
Therefore, it uses the square error as the loss function, and the output is a
set of continuous values.</p>
<p><a class="reference internal" href="generated/sklearn.neural_network.MLPRegressor.html#sklearn.neural_network.MLPRegressor" title="sklearn.neural_network.MLPRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLPRegressor</span></code></a> also supports multi-output regression, in
which a sample can have more than one target.</p>
</section>
<section id="regularization">
<h2><span class="section-number">1.17.4. </span>Regularization<a class="headerlink" href="#regularization" title="Link to this heading">#</a></h2>
<p>Both <a class="reference internal" href="generated/sklearn.neural_network.MLPRegressor.html#sklearn.neural_network.MLPRegressor" title="sklearn.neural_network.MLPRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLPRegressor</span></code></a> and <a class="reference internal" href="generated/sklearn.neural_network.MLPClassifier.html#sklearn.neural_network.MLPClassifier" title="sklearn.neural_network.MLPClassifier"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLPClassifier</span></code></a> use parameter <code class="docutils literal notranslate"><span class="pre">alpha</span></code>
for regularization (L2 regularization) term which helps in avoiding overfitting
by penalizing weights with large magnitudes. Following plot displays varying
decision function with value of alpha.</p>
<figure class="align-center">
<a class="reference external image-reference" href="../auto_examples/neural_networks/plot_mlp_alpha.html"><img alt="../_images/sphx_glr_plot_mlp_alpha_001.png" src="../_images/sphx_glr_plot_mlp_alpha_001.png" style="width: 1275.0px; height: 675.0px;" />
</a>
</figure>
<p>See the examples below for further information.</p>
<p class="rubric">Examples</p>
<ul class="simple">
<li><p><a class="reference internal" href="../auto_examples/neural_networks/plot_mlp_alpha.html#sphx-glr-auto-examples-neural-networks-plot-mlp-alpha-py"><span class="std std-ref">Varying regularization in Multi-layer Perceptron</span></a></p></li>
</ul>
</section>
<section id="algorithms">
<h2><span class="section-number">1.17.5. </span>Algorithms<a class="headerlink" href="#algorithms" title="Link to this heading">#</a></h2>
<p>MLP trains using <a class="reference external" href="https://en.wikipedia.org/wiki/Stochastic_gradient_descent">Stochastic Gradient Descent</a>,
<a class="reference external" href="https://arxiv.org/abs/1412.6980">Adam</a>, or
<a class="reference external" href="https://en.wikipedia.org/wiki/Limited-memory_BFGS">L-BFGS</a>.
Stochastic Gradient Descent (SGD) updates parameters using the gradient of the
loss function with respect to a parameter that needs adaptation, i.e.</p>
<div class="math notranslate nohighlight">
\[w \leftarrow w - \eta (\alpha \frac{\partial R(w)}{\partial w}
+ \frac{\partial Loss}{\partial w})\]</div>
<p>where <span class="math notranslate nohighlight">\(\eta\)</span> is the learning rate which controls the step-size in
the parameter space search. <span class="math notranslate nohighlight">\(Loss\)</span> is the loss function used
for the network.</p>
<p>More details can be found in the documentation of
<a class="reference external" href="https://scikit-learn.org/stable/modules/sgd.html">SGD</a></p>
<p>Adam is similar to SGD in a sense that it is a stochastic optimizer, but it can
automatically adjust the amount to update parameters based on adaptive estimates
of lower-order moments.</p>
<p>With SGD or Adam, training supports online and mini-batch learning.</p>
<p>L-BFGS is a solver that approximates the Hessian matrix which represents the
second-order partial derivative of a function. Further it approximates the
inverse of the Hessian matrix to perform parameter updates. The implementation
uses the Scipy version of <a class="reference external" href="https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.fmin_l_bfgs_b.html">L-BFGS</a>.</p>
<p>If the selected solver is ‘L-BFGS’, training does not support online nor
mini-batch learning.</p>
</section>
<section id="complexity">
<h2><span class="section-number">1.17.6. </span>Complexity<a class="headerlink" href="#complexity" title="Link to this heading">#</a></h2>
<p>Suppose there are <span class="math notranslate nohighlight">\(n\)</span> training samples, <span class="math notranslate nohighlight">\(m\)</span> features, <span class="math notranslate nohighlight">\(k\)</span>
hidden layers, each containing <span class="math notranslate nohighlight">\(h\)</span> neurons - for simplicity, and <span class="math notranslate nohighlight">\(o\)</span>
output neurons. The time complexity of backpropagation is
<span class="math notranslate nohighlight">\(O(i \cdot n \cdot (m \cdot h + (k - 1) \cdot h \cdot h + h \cdot o))\)</span>, where <span class="math notranslate nohighlight">\(i\)</span> is the number
of iterations. Since backpropagation has a high time complexity, it is advisable
to start with smaller number of hidden neurons and few hidden layers for
training.</p>
<details class="sd-sphinx-override sd-dropdown sd-card sd-mb-3" id="mathematical-formulation">
<summary class="sd-summary-title sd-card-header">
<span class="sd-summary-text">Mathematical formulation<a class="headerlink" href="#mathematical-formulation" title="Link to this dropdown">#</a></span><span class="sd-summary-state-marker sd-summary-chevron-right"><svg version="1.1" width="1.5em" height="1.5em" class="sd-octicon sd-octicon-chevron-right" viewBox="0 0 24 24" aria-hidden="true"><path d="M8.72 18.78a.75.75 0 0 1 0-1.06L14.44 12 8.72 6.28a.751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018l6.25 6.25a.75.75 0 0 1 0 1.06l-6.25 6.25a.75.75 0 0 1-1.06 0Z"></path></svg></span></summary><div class="sd-summary-content sd-card-body docutils">
<p class="sd-card-text">Given a set of training examples <span class="math notranslate nohighlight">\((x_1, y_1), (x_2, y_2), \ldots, (x_n, y_n)\)</span>
where <span class="math notranslate nohighlight">\(x_i \in \mathbf{R}^n\)</span> and <span class="math notranslate nohighlight">\(y_i \in \{0, 1\}\)</span>, a one hidden
layer one hidden neuron MLP learns the function <span class="math notranslate nohighlight">\(f(x) = W_2 g(W_1^T x + b_1) + b_2\)</span>
where <span class="math notranslate nohighlight">\(W_1 \in \mathbf{R}^m\)</span> and <span class="math notranslate nohighlight">\(W_2, b_1, b_2 \in \mathbf{R}\)</span> are
model parameters. <span class="math notranslate nohighlight">\(W_1, W_2\)</span> represent the weights of the input layer and
hidden layer, respectively; and <span class="math notranslate nohighlight">\(b_1, b_2\)</span> represent the bias added to
the hidden layer and the output layer, respectively.
<span class="math notranslate nohighlight">\(g(\cdot) : R \rightarrow R\)</span> is the activation function, set by default as
the hyperbolic tan. It is given as,</p>
<div class="math notranslate nohighlight">
\[g(z)= \frac{e^z-e^{-z}}{e^z+e^{-z}}\]</div>
<p class="sd-card-text">For binary classification, <span class="math notranslate nohighlight">\(f(x)\)</span> passes through the logistic function
<span class="math notranslate nohighlight">\(g(z)=1/(1+e^{-z})\)</span> to obtain output values between zero and one. A
threshold, set to 0.5, would assign samples of outputs larger or equal 0.5
to the positive class, and the rest to the negative class.</p>
<p class="sd-card-text">If there are more than two classes, <span class="math notranslate nohighlight">\(f(x)\)</span> itself would be a vector of
size (n_classes,). Instead of passing through logistic function, it passes
through the softmax function, which is written as,</p>
<div class="math notranslate nohighlight">
\[\text{softmax}(z)_i = \frac{\exp(z_i)}{\sum_{l=1}^k\exp(z_l)}\]</div>
<p class="sd-card-text">where <span class="math notranslate nohighlight">\(z_i\)</span> represents the <span class="math notranslate nohighlight">\(i\)</span> th element of the input to softmax,
which corresponds to class <span class="math notranslate nohighlight">\(i\)</span>, and <span class="math notranslate nohighlight">\(K\)</span> is the number of classes.
The result is a vector containing the probabilities that sample <span class="math notranslate nohighlight">\(x\)</span>
belong to each class. The output is the class with the highest probability.</p>
<p class="sd-card-text">In regression, the output remains as <span class="math notranslate nohighlight">\(f(x)\)</span>; therefore, output activation
function is just the identity function.</p>
<p class="sd-card-text">MLP uses different loss functions depending on the problem type. The loss
function for classification is Average Cross-Entropy, which in binary case is
given as,</p>
<div class="math notranslate nohighlight">
\[Loss(\hat{y},y,W) = -\dfrac{1}{n}\sum_{i=0}^n(y_i \ln {\hat{y_i}} + (1-y_i) \ln{(1-\hat{y_i})}) + \dfrac{\alpha}{2n} ||W||_2^2\]</div>
<p class="sd-card-text">where <span class="math notranslate nohighlight">\(\alpha ||W||_2^2\)</span> is an L2-regularization term (aka penalty)
that penalizes complex models; and <span class="math notranslate nohighlight">\(\alpha > 0\)</span> is a non-negative
hyperparameter that controls the magnitude of the penalty.</p>
<p class="sd-card-text">For regression, MLP uses the Mean Square Error loss function; written as,</p>
<div class="math notranslate nohighlight">
\[Loss(\hat{y},y,W) = \frac{1}{2n}\sum_{i=0}^n||\hat{y}_i - y_i ||_2^2 + \frac{\alpha}{2n} ||W||_2^2\]</div>
<p class="sd-card-text">Starting from initial random weights, multi-layer perceptron (MLP) minimizes
the loss function by repeatedly updating these weights. After computing the
loss, a backward pass propagates it from the output layer to the previous
layers, providing each weight parameter with an update value meant to decrease
the loss.</p>
<p class="sd-card-text">In gradient descent, the gradient <span class="math notranslate nohighlight">\(\nabla Loss_{W}\)</span> of the loss with respect
to the weights is computed and deducted from <span class="math notranslate nohighlight">\(W\)</span>.
More formally, this is expressed as,</p>
<div class="math notranslate nohighlight">
\[W^{i+1} = W^i - \epsilon \nabla {Loss}_{W}^{i}\]</div>
<p class="sd-card-text">where <span class="math notranslate nohighlight">\(i\)</span> is the iteration step, and <span class="math notranslate nohighlight">\(\epsilon\)</span> is the learning rate
with a value larger than 0.</p>
<p class="sd-card-text">The algorithm stops when it reaches a preset maximum number of iterations; or
when the improvement in loss is below a certain, small number.</p>
</div>
</details></section>
<section id="tips-on-practical-use">
<span id="mlp-tips"></span><h2><span class="section-number">1.17.7. </span>Tips on Practical Use<a class="headerlink" href="#tips-on-practical-use" title="Link to this heading">#</a></h2>
<ul>
<li><p>Multi-layer Perceptron is sensitive to feature scaling, so it
is highly recommended to scale your data. For example, scale each
attribute on the input vector X to [0, 1] or [-1, +1], or standardize
it to have mean 0 and variance 1. Note that you must apply the <em>same</em>
scaling to the test set for meaningful results.
You can use <a class="reference internal" href="generated/sklearn.preprocessing.StandardScaler.html#sklearn.preprocessing.StandardScaler" title="sklearn.preprocessing.StandardScaler"><code class="xref py py-class docutils literal notranslate"><span class="pre">StandardScaler</span></code></a> for standardization.</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.preprocessing</span> <span class="kn">import</span> <span class="n">StandardScaler</span>
<span class="gp">>>> </span><span class="n">scaler</span> <span class="o">=</span> <span class="n">StandardScaler</span><span class="p">()</span>
<span class="gp">>>> </span><span class="c1"># Don't cheat - fit only on training data</span>
<span class="gp">>>> </span><span class="n">scaler</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">X_train</span> <span class="o">=</span> <span class="n">scaler</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X_train</span><span class="p">)</span>
<span class="gp">>>> </span><span class="c1"># apply same transformation to test data</span>
<span class="gp">>>> </span><span class="n">X_test</span> <span class="o">=</span> <span class="n">scaler</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X_test</span><span class="p">)</span>
</pre></div>
</div>
<p>An alternative and recommended approach is to use
<a class="reference internal" href="generated/sklearn.preprocessing.StandardScaler.html#sklearn.preprocessing.StandardScaler" title="sklearn.preprocessing.StandardScaler"><code class="xref py py-class docutils literal notranslate"><span class="pre">StandardScaler</span></code></a> in a
<a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a></p>
</li>
<li><p>Finding a reasonable regularization parameter <span class="math notranslate nohighlight">\(\alpha\)</span> is best done
using <a class="reference internal" href="generated/sklearn.model_selection.GridSearchCV.html#sklearn.model_selection.GridSearchCV" title="sklearn.model_selection.GridSearchCV"><code class="xref py py-class docutils literal notranslate"><span class="pre">GridSearchCV</span></code></a>, usually in the range
<code class="docutils literal notranslate"><span class="pre">10.0</span> <span class="pre">**</span> <span class="pre">-np.arange(1,</span> <span class="pre">7)</span></code>.</p></li>
<li><p>Empirically, we observed that <code class="docutils literal notranslate"><span class="pre">L-BFGS</span></code> converges faster and
with better solutions on small datasets. For relatively large
datasets, however, <code class="docutils literal notranslate"><span class="pre">Adam</span></code> is very robust. It usually converges
quickly and gives pretty good performance. <code class="docutils literal notranslate"><span class="pre">SGD</span></code> with momentum or
nesterov’s momentum, on the other hand, can perform better than
those two algorithms if learning rate is correctly tuned.</p></li>
</ul>
</section>
<section id="more-control-with-warm-start">
<h2><span class="section-number">1.17.8. </span>More control with warm_start<a class="headerlink" href="#more-control-with-warm-start" title="Link to this heading">#</a></h2>
<p>If you want more control over stopping criteria or learning rate in SGD,
or want to do additional monitoring, using <code class="docutils literal notranslate"><span class="pre">warm_start=True</span></code> and
<code class="docutils literal notranslate"><span class="pre">max_iter=1</span></code> and iterating yourself can be helpful:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">X</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span>
<span class="gp">>>> </span><span class="n">y</span> <span class="o">=</span> <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">1</span><span class="p">]</span>
<span class="gp">>>> </span><span class="n">clf</span> <span class="o">=</span> <span class="n">MLPClassifier</span><span class="p">(</span><span class="n">hidden_layer_sizes</span><span class="o">=</span><span class="p">(</span><span class="mi">15</span><span class="p">,),</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">max_iter</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">warm_start</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="gp">>>> </span><span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="mi">10</span><span class="p">):</span>
<span class="gp">... </span> <span class="n">clf</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span>
<span class="gp">... </span> <span class="c1"># additional monitoring / inspection</span>
<span class="go">MLPClassifier(...</span>
</pre></div>
</div>
<details class="sd-sphinx-override sd-dropdown sd-card sd-mb-3" id="references">
<summary class="sd-summary-title sd-card-header">
<span class="sd-summary-text">References<a class="headerlink" href="#references" title="Link to this dropdown">#</a></span><span class="sd-summary-state-marker sd-summary-chevron-right"><svg version="1.1" width="1.5em" height="1.5em" class="sd-octicon sd-octicon-chevron-right" viewBox="0 0 24 24" aria-hidden="true"><path d="M8.72 18.78a.75.75 0 0 1 0-1.06L14.44 12 8.72 6.28a.751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018l6.25 6.25a.75.75 0 0 1 0 1.06l-6.25 6.25a.75.75 0 0 1-1.06 0Z"></path></svg></span></summary><div class="sd-summary-content sd-card-body docutils">
<ul class="simple">
<li><p class="sd-card-text"><a class="reference external" href="https://www.iro.umontreal.ca/~pift6266/A06/refs/backprop_old.pdf">“Learning representations by back-propagating errors.”</a>
Rumelhart, David E., Geoffrey E. Hinton, and Ronald J. Williams.</p></li>
<li><p class="sd-card-text"><a class="reference external" href="https://leon.bottou.org/projects/sgd">“Stochastic Gradient Descent”</a> L. Bottou - Website, 2010.</p></li>
<li><p class="sd-card-text"><a class="reference external" href="http://ufldl.stanford.edu/wiki/index.php/Backpropagation_Algorithm">“Backpropagation”</a>
Andrew Ng, Jiquan Ngiam, Chuan Yu Foo, Yifan Mai, Caroline Suen - Website, 2011.</p></li>
<li><p class="sd-card-text"><a class="reference external" href="http://yann.lecun.com/exdb/publis/pdf/lecun-98b.pdf">“Efficient BackProp”</a>
Y. LeCun, L. Bottou, G. Orr, K. Müller - In Neural Networks: Tricks of the Trade 1998.</p></li>