Multi-classes Semi-supervised Learning on Riemannian Manifolds

被引：0

作者：

Zhao, Zhong-Qiu ^{[1
,2
]}

Glotin, Herve ^{[2
]}

Gao, Jun ^{[1
]}

Wu, Xin-Dong ^{[1
]}

机构：

[1] Hefei Univ Technol, Comp & Informat Sch, Hefei 230009, Anhui, Peoples R China

[2] Univ Sud Toulon Var, CNRS, LSIS, UMR, Toulon, France

来源：

PROCEEDINGS OF THE 2009 INTERNATIONAL CONFERENCE ON COMPUTATIONAL INTELLIGENCE AND NATURAL COMPUTING, VOL I | 2009年

基金：

中国国家自然科学基金;

关键词：

semi-supervised learning; Riemannian manifolds; learning speed; multi-classes problem; DIMENSIONALITY REDUCTION;

D O I：

10.1109/CINC.2009.105

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

A modified semi-supervised learning algorithm based on Riemannian manifolds is proposed, which extends the applicable field to the case of multi-classes problem. Furthermore, the modified version largely increases the learning speed, while attains the classification performance as satisfying as the original algorithm.

引用

页码：527 / +

页数：3