A Granular Evolutionary Algorithm Based on Cultural Evolution

摘要

Analogous to biological evolution, cultural evolution also is a kind of optimal mechanism of nature. Studying this mechanism might possibly provide a more efficient computation for solving complicated problems, such as knowledge acquisition in large data set. In this paper, an algorithm, granular evolutionary algorithm for data classification, simply written as GEA, is proposed based on cultural evolution and granular computing. The proposed algorithm is essentially a granular computation, which is characterized by computing with granules. Each granule consists of some individuals, which itself also is an evolutionary population. The algorithm is realized in PVM environment by agent technology, and the experimental results certify its validity. Further analysis can find that the proposed algorithm has relatively better performance from large data sets.