In pattern classification system, many irrelevant and redundant features will lessen the performance of classifiers.
Therefore, a feature subset selection method based on asynchronous parallel particle swarm optimization algorithm is
proposed. This algorithm uses the binary particle swarm optimization to select feature subsets, and takes advantage of
asynchronous parallel strategy to enhance time efficiency. In order to balance effectually the global exploration and the
local exploitation of swarm, an uniform chaos mutation also is proposed by making the best use of the ergodicity, stochastic
property and regularity of chaos. By compared with four known feature selection methods, the results show the effectiveness
of the proposed algorithm.