An ICA with reference based on artificial fish swarm algorithm

Abstract

The independent component analysis with reference algorithm uses gradient method to optimize the cost function, this makes it easily fall into local optimal solution. To overcome the problem, this paper proposes a new independent component analysis with reference algorithm that has global convergence. The new algorithm uses artificial fish swarm algorithm with global convergence to optimize cost function of independent component analysis algorithm with reference. It accords to the behavior of artificial fish preying, swarming, following and food consistence to update artificial fish position, which is to update the separation matrix and research the separation matrix optimal solution of independent component analysis algorithm with reference. Comparing with the original algorithm based on gradient method, the new algorithm does not need to calculate the gradient of cost function and has higher accuracy. Simulation results show that the new method is effective.