Fuzzy classifier design with coevolutionary algorithms applying for speaker identification

The problem of speaker identification is considered in this article. Classification problem Japanese vowels» from UCI repository is used as source data. This problem was solved with a fuzzy classifier as a classification method that is able to extract cause-and-effect relations from source data. A new method of fuzzy classifier rule base design with coevolutionary algorithms was applied. It is multistep fuzzy classifier design based on multiple repetition of previous fuzzy classifier design with self-tuning coevolutionary algorithms. Computational investigation of fuzzy classifier design with coevolutionary algorithms for different numbers of speakers and for different number of the used fuzzy rules was performed. The proposed method allows getting acceptable classification efficiency for a test sample: from 0.985 for two speakers to 0.786 for nine speakers.