Knowledge extraction from a class of support vector machines using the fuzzy all-permutations rule-base
Shahaf Duenyas, Michael Margaliot · 2011
Support vector machines (SVMs) proved to be highly efficient in various classification tasks. However, the knowledge learned by the SVM is encoded in a long list of parameter values and it is not easy to comprehend what the SVM is actually computing. We show that certain types of SVMs are mathematically equivalent to a specific fuzzy — rule base, the fuzzy all — permutations rule base (FARB). This equivalence can be used to provide a symbolic representation of the SVM functioning. This leads to a new approach for knowledge extraction from SVMs. Two simple examples demonstrate the effectiveness of this approach.