×

zbMATH — the first resource for mathematics

Exact simplification of support vector solutions. (English) Zbl 1037.68111
Summary: This paper demonstrates that standard algorithms for training support vector machines generally produce solutions with a greater number of support vectors than are strictly necessary. An algorithm is presented that allows unnecessary support vectors to be recognized and eliminated while leaving the solution otherwise unchanged. The algorithm is applied to a variety of benchmark data sets (for both classification and regression) and in most cases the procedure leads to a reduction in the number of support vectors. In some cases the reduction is substantial.

MSC:
68T05 Learning and adaptive systems in artificial intelligence
Software:
SVMTorch
PDF BibTeX XML Cite
Full Text: DOI