L1-NORM HIGHER-ORDER ORTHOGONAL ITERATIONS FOR ROBUST TENSOR ANALYSIS

被引：0

作者：

Chachlakis, Dimitris G. ^{[1
]}

Prater-Bennette, Ashley ^{[2
]}

Markopoulos, Panos P. ^{[1
]}

机构：

[1] Rochester Inst Technol, Dept Elect & Microelect Engn, Rochester, NY 14623 USA

[2] Air Force Res Lab, Rome, NY 13441 USA

来源：

2020 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING | 2020年

基金：

美国国家科学基金会;

关键词：

D O I：

10.1109/icassp40776.2020.9053701

中图分类号：

O42 [声学];

学科分类号：

070206 ; 082403 ;

摘要：

Standard Tucker tensor decomposition seeks to maximize the L2-norm of the compressed tensor; thus, it is very responsive to outlying/high-magnitude entries among the processed data. To counteract the impact of outliers in tensor data analysis, we propose L1-Tucker: a reformulation of standard Tucker decomposition, resulting by simple substitution of the outlier-responsive L2-norm by the sturdier L1-norm. Then, we propose the L1-norm Higher Order Orthogonal Iterations (L1-HOOI) algorithm for the approximate solution to L1-Tucker. Our numerical studies on data reconstruction and classification corroborate that L1-HOOI exhibits sturdy resistance against outliers compared to standard counterparts.

引用

页码：4826 / 4830

页数：5

共 28 条

[11]

Golub G H., 2013, Matrix Computations, V4th edition

[12]

Gower J.C., 2004, Procrustes problems

[13] Tensor Decompositions and Applications [J].

Kolda, Tamara G. ;

Bader, Brett W. .

SIAM REVIEW, 2009, 51 (03) :455-500

[14] Compressed-Sensed-Domain L1-PCA Video Surveillance [J].

Liu, Ying ;

Pados, Dimitris A. .

IEEE TRANSACTIONS ON MULTIMEDIA, 2016, 18 (03) :351-363

[15] Tensor Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Tensors via Convex Optimization [J].

Lu, Canyi ;

Feng, Jiashi ;

Chen, Yudong ;

Liu, Wei ;

Lin, Zhouchen ;

Yan, Shuicheng .

2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2016, :5249-5257

[16]

Markopoulos PP, 2018, IEEE GLOB CONF SIG, P1353, DOI 10.1109/GlobalSIP.2018.8646385

[17] The Exact Solution to Rank-1 L1-Norm TUCKER2 Decomposition [J].

Markopoulos, Panos P. ;

Chachlakis, Dimitris G. ;

Papalexakis, Evangelos E. .

IEEE SIGNAL PROCESSING LETTERS, 2018, 25 (04) :511-515

[18] Efficient L1-Norm Principal-Component Analysis via Bit Flipping [J].

Markopoulos, Panos P. ;

Kundu, Sandipan ;

Chamadia, Shubham ;

Pados, Dimitris A. .

IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2017, 65 (16) :4252-4264

[19]

Markopoulos PP, 2017, IEEE RAD CONF, P1807, DOI 10.1109/RADAR.2017.7944501

[20]

Markopoulos PP, 2015, IEEE IMAGE PROC, P1225, DOI 10.1109/ICIP.2015.7350995

← 1 2 3 →