Blind source separation based on improved natural gradient algorithm
暂无分享,去创建一个
The natural gradient algorithm is the most basic independent component analysis (ICA) algorithm. Because the traditional natural gradient algorithm adopts fixed-step-size, the choice of step size directly affects the convergence speed and steady-state performance. This paper proposes an improved natural gradient algorithm by using the difference between the separation matrixes to control the factor of step size. The algorithm is a good solution to the trade-offs problems of convergence speed and steady-state performance. Meanwhile, the complexity of the algorithm is lower than the algorithm of reference [2] and reference [11]. The computer simulations have proved the effectiveness of the algorithm.
[1] Terrence J. Sejnowski,et al. An Information-Maximization Approach to Blind Separation and Blind Deconvolution , 1995, Neural Computation.
[2] Andrzej Cichocki,et al. Robust neural networks with on-line learning for blind identification and blind separation of sources , 1996 .
[3] Thomas P. von Hoff,et al. STEP-SIZE CONTROL IN BLIND SOURCE SEPARATION , 2000 .
[4] Shun-ichi Amari,et al. Natural Gradient Works Efficiently in Learning , 1998, Neural Computation.