Incremental training of support vector machines

被引:134
作者
Shilton, A [1 ]
Palaniswami, M
Ralph, D
Tsoi, AC
机构
[1] Univ Melbourne, Dept Elect & Elect Engn, Ctr Expertise Networked Decis & Sensor Syst, Melbourne, Vic 3010, Australia
[2] Univ Cambridge, Judge Inst Management Studies, Cambridge CB2 1AG, England
[3] Univ Wollongong, Wollongong, NSW 2522, Australia
来源
IEEE TRANSACTIONS ON NEURAL NETWORKS | 2005年 / 16卷 / 01期
关键词
active set method; incremental training; quadratic programming; support vector machines (SVMs); warm start algorithm;
D O I
10.1109/TNN.2004.836201
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We propose a new algorithm for the incremental training of support vector machines (SVMs) that is suitable for problems of sequentially arriving data and fast constraint parameter variation. Our method involves using a "warm-start" algorithm for the training of SVMs, which allows us to take advantage of the natural incremental properties of the standard active set approach to linearly constrained optimization problems. Incremental training involves quickly retraining a support vector machine after adding a small number of additional training vectors to the training set of an existing (trained) support vector machine. Similarly, the problem of fast constraint parameter variation involves quickly retraining an existing support vector machine using the same training set but different constraint parameters. In both cases, we demonstrate the computational superiority of incremental training over the usual batch retraining method.
引用
收藏
页码:114 / 131
页数:18
相关论文
共 17 条
[1]  
[Anonymous], 1981, PRACTICAL METHODS OP
[2]  
Blake C.L., 1998, UCI repository of machine learning databases
[3]   A tutorial on Support Vector Machines for pattern recognition [J].
Burges, CJC .
DATA MINING AND KNOWLEDGE DISCOVERY, 1998, 2 (02) :121-167
[4]  
Cauwenberghs G, 2001, ADV NEUR IN, V13, P409
[5]   Choosing multiple parameters for support vector machines [J].
Chapelle, O ;
Vapnik, V ;
Bousquet, O ;
Mukherjee, S .
MACHINE LEARNING, 2002, 46 (1-3) :131-159
[6]   A DIRECT ACTIVE SET ALGORITHM FOR LARGE SPARSE QUADRATIC PROGRAMS WITH SIMPLE BOUNDS [J].
COLEMAN, TF ;
HULBERT, LA .
MATHEMATICAL PROGRAMMING, 1989, 45 (03) :373-406
[7]  
CORTES C, 1995, MACH LEARN, V20, P273, DOI 10.1023/A:1022627411411
[8]  
FINE S, P NIPS2000
[9]  
FINE S, 2002, ADV NEURAL INFORMATI
[10]  
GILL PE, 1974, MATH COMPUT, V28, P505, DOI 10.1090/S0025-5718-1974-0343558-6