Unsupervised discovery of invariances

被引：5

作者：

Eglen, S ^{[1
]}

Bray, A ^{[1
]}

Stone, J ^{[1
]}

机构：

[1] UNIV SUSSEX,SCH COGNIT & COMP SCI,BRIGHTON BN1 9QH,E SUSSEX,ENGLAND

来源：

NETWORK-COMPUTATION IN NEURAL SYSTEMS | 1997年 / 8卷 / 04期

关键词：

D O I：

10.1088/0954-898X/8/4/006

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

The grey level profiles of adjacent image regions tend to be different, whilst the 'hidden' physical parameters associated with these regions (e.g. surface depth, edge orientation) tend to have similar values. We demonstrate that a network in which adjacent units receive inputs from adjacent image regions learns to code for hidden parameters. The learning rule takes advantage of the spatial smoothness of physical parameters in general to discover particular parameters embedded in grey level profiles which vary rapidly across an input image. We provide examples in which networks discover stereo disparity and feature orientation as invariances underlying image data.

引用

页码：441 / 452

页数：12

共 15 条

[1] SELF-ORGANIZING NEURAL NETWORK THAT DISCOVERS SURFACES IN RANDOM-DOT STEREOGRAMS [J].

BECKER, S ;

HINTON, GE .

NATURE, 1992, 355 (6356) :161-163

[2] A FUNCTIONAL MICROCIRCUIT FOR CAT VISUAL-CORTEX [J].

DOUGLAS, RJ ;

MARTIN, KAC .

JOURNAL OF PHYSIOLOGY-LONDON, 1991, 440 :735-769

[3]

EGLEN SJ, 1996, 404 CSRP U SUSS SCH

[4] Learning Invariance from Transformation Sequences [J].

Foldiak, Peter .

NEURAL COMPUTATION, 1991, 3 (02) :194-200

[5]

Horn B., 1986, Robot Vision

[6]

LINSKER R, 1988, IEEE COMPUT, V21, P105, DOI DOI 10.1109/2.36

[7] Learning generalisation and localisation: Competition for stimulus type and receptive field [J].

Oram, MW ;

Foldiak, P .

NEUROCOMPUTING, 1996, 11 (2-4) :297-321

[8] THE DISCOVERY OF STRUCTURE BY MULTI-STREAM NETWORKS OF LOCAL PROCESSORS WITH CONTEXTUAL GUIDANCE [J].

PHILLIPS, WA ;

KAY, J ;

SMYTH, D .

NETWORK-COMPUTATION IN NEURAL SYSTEMS, 1995, 6 (02) :225-246

[9] LEARNING REPRESENTATIONS BY BACK-PROPAGATING ERRORS [J].

RUMELHART, DE ;

HINTON, GE ;

WILLIAMS, RJ .

NATURE, 1986, 323 (6088) :533-536

[10]

SCHRAUDOLPH NN, 1992, ADV NEUR IN, V4, P1017

← 1 2 →