zbMATH — the first resource for mathematics

Learning by kernel polarization. (English) Zbl 1087.68102
Summary: Kernels are key components of pattern recognition mechanisms. We propose a universal kernel optimality criterion, which is independent of the classifier to be used. Defining data polarization as a process by which points of different classes are driven to geometrically opposite locations in a confined domain, we propose selecting the kernel parameter values that polarize the data in the associated feature space. Conversely, the kernel is said to be polarized by the data. Kernel polarization gives rise to an unconstrained optimization problem. We show that complete kernel polarization yields consistent classification by kernel-sum classifiers. Tested on real-life data, polarized kernels demonstrate a clear advantage over the Euclidean distance in proximity classifiers. Embedded in a support vectors classifier, kernel polarization is found to yield about the same performance as exhaustive parameter search.

68T10 Pattern recognition, speech recognition
68T05 Learning and adaptive systems in artificial intelligence
Full Text: DOI
[1] DOI: 10.1016/0925-2312(96)00047-1 · Zbl 05472638
[2] DOI: 10.1023/A:1012450327387 · Zbl 0998.68101
[3] DOI: 10.1109/TIT.1967.1053964 · Zbl 0154.44505
[4] DOI: 10.1214/aoms/1177704472 · Zbl 0116.11302
[5] DOI: 10.1037/h0042519
[6] Vapnik V., Dokl. Akad. Nauk SSSR 181 pp 915– (1968)
[7] DOI: 10.1137/1116025 · Zbl 0247.60005
[8] Vapnik V., Automation and Remote Control 24 pp 774– (1963)
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.