Application of Genetic Algorithm to Independent Component Analysis

Abstract

In this paper,application of GA to ICA is studied.The algorithm takes fully advantage of features in separated signals.Not only does it overcome usual gradient algorithms'disadvantages of falling into partial optimisation,but also it avoids complete random search in genetic algorithms.This algorithm takes KL divergence as its original objective function and forms a constrained multi-objective optimization algorithm by subjecting to separated matrix.And it solves the problem by iteration,and constructs a new fitness function based on the original objective function.Then it takes float coding,arithmetic combinatorial crossover,mutation along the weighted gradient direction and selection by the roulette strategy.After a few iterative steps,the optima is obtained by using mutation,crossover and selection operators.Compared with other usual ICA algorithms,simulations show that this new algorithm has better performance and separation.

Cite this paper

@article{ouyang2010application,
  title = {{Application of Genetic Algorithm to Independent Component Analysis}},
  author = {Ouyang, Baiping},
  journal = {Journal of Hebei North University},
  year = {2010}
}

Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.

Related papers