GENETIC APPROACH TO SUPPORT SETS SYSTEM ETIMATION FOR ALVOT

摘要

In this paper we propose an alternative method, based on genetic algorithms, to estimate the support sets system for the supervised classification model ALVOT. Usually, this system is taken as the set of all typical testers, but this has a problem: algorithms for calculating all typical tester have very high complexity. Also, in some practical cases, the number of typical testors can be too large. This can become typical testors inapplicable for the classification stage. The proposed method allows generating support sets systems of limited size, but with a high efficiency for classification. Some performance examples, of the new method, are exposed. Quality for classification, of the results, is compared against typical testors option.