An error-entropy minimization algorithm for supervised training of nonlinear adaptive systems

摘要: The paper investigates error-entropy-minimization in adaptive systems training. We prove the equivalence between minimization of error's Renyi (1970) entropy order /spl alpha/ and a Csiszar (1981) distance measure densities desired system outputs. A nonparametric estimator for Renyi's is presented, it shown that global minimum this same as actual entropy. performance criterion compared with mean-square-error-minimization short-term prediction chaotic time series nonlinear identification.