On-line learning in changing environments with applications in supervised and unsupervised learning

被引：43

作者：

Murata, N

Kawanabe, M

Ziehe, A

Müller, KR

Amari, S

机构：

[1] Fraunhofer FIRST, D-12489 Berlin, Germany

[2] Waseda Univ, Sch Sci & Engn, Tokyo 169, Japan

[3] Univ Potsdam, Dept Comp Sci, D-14482 Potsdam, Germany

[4] RIKEN, Brain Sci Inst, Wako, Saitama 35101, Japan

来源：

NEURAL NETWORKS | 2002年 / 15卷 / 4-6期

关键词：

on-line learning; adaptive learning rate; ICA; blind source separation; stochastic gradient descent; supervised learning; unsupervised learning;

D O I：

10.1016/S0893-6080(02)00060-6

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

An adaptive on-line algorithm extending the learning of learning idea is proposed and theoretically motivated. Relying only on gradient flow information it can be applied to learning continuous functions or distributions, even when no explicit loss function is given and the Hessian is not available. The framework is applied for unsupervised and supervised learning. Its efficiency is demonstrated for drifting and switching non-stationary blind separation tasks of acoustic signals. Furthermore applications to classification (US postal service data set) and time-series prediction in changing environments are presented. (C) 2002 Elsevier Science Ltd. All rights reserved.

引用

页码：743 / 760

页数：18

共 49 条

[1]

Almeida Luis B, 1998, ON LINE LEARNING NEU, P111

[2]

Amari S, 1996, ADV NEUR IN, V8, P757

[3] Natural gradient works efficiently in learning [J].