Recursive finite Newton algorithm for support vector regression in the primal. (English) Zbl 1118.68114

Bo, Liefeng; Wang, Ling; Jiao, Licheng

Neural Comput. 19, No. 4, 1082-1096 (2007).

Summary: Some algorithms in the primal have been recently proposed for training support vector machines. This letter follows those studies and develops a recursive finite Newton algorithm (IHLF-SVR-RFN) for training nonlinear support vector regression. The insensitive Huber loss function and the computation of the Newton step are discussed in detail. Comparisons with LIBSVM 2.82 show that the proposed algorithm gives promising results.

MSC:

68T05	Learning and adaptive systems in artificial intelligence
62M10	Time series, auto-correlation, regression, etc. in statistics (GARCH)
62M45	Neural nets and related approaches to inference from stochastic processes

Keywords:

nonlinear support vector regression

Software:

LIBSVM

References:

[1]	DOI: 10.1023/A:1009715923555 · Zbl 05470543
[2]	Fan R. E., Journal of Machine Learning Research 6 pp 1889– (2005)
[3]	DOI: 10.1016/S0925-2312(03)00379-5 · Zbl 02060367
[4]	DOI: 10.1007/BF01442169 · Zbl 0542.49011
[5]	Keerthi S. S., Journal of Machine Learning Research 7 pp 1493– (2006)
[6]	Keerthi S. S., Journal of Machine Learning Research 6 pp 341– (2005)
[7]	DOI: 10.1162/089976601300014493 · Zbl 1085.68629
[8]	DOI: 10.1214/aoms/1177697089 · Zbl 0193.45201
[9]	DOI: 10.1007/BF01933216 · Zbl 0717.65118
[10]	DOI: 10.1080/1055678021000028375 · Zbl 1065.90078
[11]	DOI: 10.1109/72.870050
[12]	DOI: 10.1023/B:STCO.0000035301.49549.88