×

Feature selection via coalitional game theory. (English) Zbl 1173.91305

Summary: We present and study the contribution-selection algorithm (CSA), a novel algorithm for feature selection. The algorithm is based on the multiperturbation shapley analysis (MSA), a framework that relies on game theory to estimate usefulness. The algorithm iteratively estimates the usefulness of features and selects them accordingly, using either forward selection or backward elimination. It can optimize various performance measures over unseen data such as accuracy, balanced error rate, and area under receiver-operator-characteristic curve. Empirical comparison with several other existing feature selection methods shows that the backward elimination variant of CSA leads to the most accurate classification results on an array of data sets.

MSC:

91A12 Cooperative games
PDF BibTeX XML Cite
Full Text: DOI Link

References:

[1] DOI: 10.1073/pnas.96.12.6745
[2] DOI: 10.1287/opre.26.6.956 · Zbl 0417.90059
[3] DOI: 10.1016/S0004-3702(97)00063-5 · Zbl 0904.68142
[4] DOI: 10.1023/A:1010933404324 · Zbl 1007.68152
[5] DOI: 10.1006/jcss.2001.1754 · Zbl 0996.68026
[6] DOI: 10.1016/S0895-4356(98)00002-X
[7] Gillick L., ICASSP 1 pp 532– (1989)
[8] DOI: 10.1162/153244303322753616 · Zbl 1102.68556
[9] DOI: 10.1023/A:1012487302797 · Zbl 0998.68111
[10] DOI: 10.1162/0899766041336387 · Zbl 1086.68570
[11] DOI: 10.1162/artl.2006.12.3.333
[12] DOI: 10.1016/S0004-3702(97)00043-X · Zbl 0904.68143
[13] DOI: 10.1162/153244303322753698 · Zbl 1102.68578
[14] DOI: 10.1162/153244303322753724 · Zbl 1102.68586
[15] DOI: 10.2307/1951053
[16] DOI: 10.1287/mnsc.8.3.325 · Zbl 0995.90578
[17] DOI: 10.1162/153244303322753733 · Zbl 1102.68598
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.