Tensor-Train decomposition for image recognition

被引：5

作者：

Brandoni, D. ^{[1
]}

Simoncini, V ^{[1
,2
,3
]}

机构：

[1] Univ Bologna, Dipartimento Matemat, Piazza Porta San Donato 5, Bologna, Italy

[2] Univ Bologna, AM2, Piazza Porta San Donato 5, Bologna, Italy

[3] CNR, IMATI, Pavia, Italy

来源：

CALCOLO | 2020年 / 57卷 / 01期

关键词：

Tensor-Train; Tensor decomposition; HOSVD; Image classification; Face recognition;

D O I：

10.1007/s10092-020-0358-8

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

We explore the potential of Tensor-Train (TT) decompositions in the context of multi-feature face or object recognition strategies. We devise a new recognition algorithm that can handle three or more way tensors in the TT format, and propose a truncation strategy to limit memory usage. Numerical comparisons with other related methods-including the well established recognition algorithm based on high-order SVD-illustrate the features of the various strategies on benchmark datasets.

引用

页数：24

共 25 条

[1]

[Anonymous], ECML PKDD WORKSH

[2]

[Anonymous], 2008, Recent Advances in Face Recognition

[3]

[Anonymous], ARXIV150906569

[4]

[Anonymous], ARXIV160909230

[5]

[Anonymous], 2013, Matrix Computations

[6]

[Anonymous], ARXIV14034462

[7]

[Anonymous], 2017, ARXIV170701786

[8]

Bader B. W., 2017, MATLAB tensor toolbox version 3.0-dev

[9]

BOUSSELHAM M, 2017, 2017 INT C SOFT COMP, P1, DOI DOI 10.1109/ICSOFTCOMP.2017.8280084

[10] Selecting among three-mode principal component models of different types and complexities: A numerical convex hull based method [J].

Ceulemans, E ;

Kiers, HAL .

BRITISH JOURNAL OF MATHEMATICAL & STATISTICAL PSYCHOLOGY, 2006, 59 :133-150

← 1 2 3 →