An Improved Particle Swarm Optimization Application to Independent Component Analysis

The traditional searching scheme of independent component analysis (ICA) is based on gradient algorithm. And a learning step size is required beforehand. It couldn´t resolve the problem of convergence. To overcome the drawback, an improved particle swarm optimization (PSO) is applied to ICA algorithm. Firstly, the dynamic inertia weight which is based on evolution speed and aggregation degree is introduced into PSO. And then, based on the analysis of ICA, a fitness function of PSO was defined. Finally, the detailed algorithm was given by using improved PSO. Based on TIMIT corpus and Noise-92 database, the experiments were implemented. The results indicate that the performance of DPSO-ICA algorithm is superior to the traditional FastICA for processing mixed noisy speech signals.