A chaotic prediction algorithm using a new cost function
The traditional cost function,minimization mean square prediction error is not a proper cost function in chaotic series prediction,for many chaotic signals are non-Gaussian distributions.Then we present using minimization error negentropy as new cost function,and derive the nonlinear approximation method.In simulation,the algorithm shows an enhanced performance to a common two order Volterra prediction.
Chaotic time series density function cost function negentropy
Bu Yun Kang Wan Xin
School of Electrical and Information Engineering Xihua University Chengdu,China
国际会议
杭州
英文
494-496
2013-03-22（万方平台首次上网日期，不代表论文的发表时间）