Invariant Scattering Convolution Networks

被引:1100
作者
Bruna, Joan [1 ]
Mallat, Stephane [2 ]
机构
[1] NYU, Courant Inst, New York, NY 10003 USA
[2] Ecole Normale Super, F-75005 Paris, France
关键词
Classification; convolution networks; deformations; invariants; wavelets; MODELS;
D O I
10.1109/TPAMI.2012.230
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
A wavelet scattering network computes a translation invariant image representation which is stable to deformations and preserves high-frequency information for classification. It cascades wavelet transform convolutions with nonlinear modulus and averaging operators. The first network layer outputs SIFT-type descriptors, whereas the next layers provide complementary invariant information that improves classification. The mathematical analysis of wavelet scattering networks explains important properties of deep convolution networks for classification. A scattering representation of stationary processes incorporates higher order moments and can thus discriminate textures having the same Fourier power spectrum. State-of-the-art classification results are obtained for handwritten digits and texture discrimination, with a Gaussian kernel SVM and a generative PCA classifier.
引用
收藏
页码:1872 / 1886
页数:15
相关论文
共 37 条
[1]  
Allassonnière S, 2007, J R STAT SOC B, V69, P3
[2]   POP:: Patchwork of parts models for object recognition [J].
Amit, Yali ;
Trouve, Alain .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 2007, 75 (02) :267-282
[3]  
Anden J., IEEE T SIGN IN PRESS
[4]  
[Anonymous], ANN STAT
[5]  
[Anonymous], 1997, Festschrift for lucien le cam
[6]  
[Anonymous], P IEEE INT S CIRC SY
[7]  
[Anonymous], ARXIV12060102
[8]  
[Anonymous], P EUR S ART NEUR NET
[9]   Learning Mid-Level Features For Recognition [J].
Boureau, Y-Lan ;
Bach, Francis ;
LeCun, Yann ;
Ponce, Jean .
2010 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2010, :2559-2566
[10]  
Bouvrie J., 2009, P ADV NEUR INF PROC