Generalized information potential criterion for adaptive system training
Tóm tắt
We have previously proposed the quadratic Renyi's error entropy as an alternative cost function for supervised adaptive system training. An entropy criterion instructs the minimization of the average information content of the error signal rather than merely trying to minimize its energy. In this paper, we propose a generalization of the error entropy criterion that enables the use of any order of Renyi's entropy and any suitable kernel function in density estimation. It is shown that the proposed entropy estimator preserves the global minimum of actual entropy. The equivalence between global optimization by convolution smoothing and the convolution by the kernel in Parzen windowing is also discussed. Simulation results are presented for time-series prediction and classification where experimental demonstration of all the theoretical concepts is presented.
Từ khóa
#Adaptive systems #Entropy #Kernel #Signal processing #Cost function #Convolution #Mutual information #Source separation #Feature extraction #ChaosTài liệu tham khảo
kuo, 1993, Nonlinear dynamic modeling with artificial neural networks
kaplan, 1995, Understanding Nonlinear Dynamics, 10.1007/978-1-4612-0823-5
10.1214/aoms/1177704472
principe, 2000, information theoretic learning, Unsupervised Adaptive Filtering, i, 265
10.1109/97.923043
xu, 0, a novel measure for independent component analysis (ica), Proc ICASSP 98, ii, 1161
torkkola, 2000, mutual information in learning feature transformations, Proc 7th Int Conf Machine Learning
10.1007/978-1-4757-2440-0
10.1103/PhysRevE.53.2169
10.1103/PhysRevLett.50.346
erdogmus, 2000, comparison of entropy and mean square error criteria in adaptive system training using higher order statistics, Proc Independent Components Analysis (ICA)
10.1109/TSP.2002.1011217
10.1002/j.1538-7305.1928.tb01236.x
rubinstein, 1981, Simulation and the Monte Carlo Method, 10.1002/9780470316511
aarts, 1989, Simulated Annealing and Boltzmann Machine A Stochastic Approach to Combinatorial Optimization with Neural Computing
kapur, 1992, Entropy Optimization Principles and Applications
amari, 1985, Differential-Geometrical Methods in Statistics, 10.1007/978-1-4612-5056-2
edmonson, 1996, a global least square algorithm for adaptive iir filtering, IEEE Trans IEEE Trans Circuits Syst, 45, 379
10.1002/j.1538-7305.1948.tb01338.x
10.1016/0165-1684(94)90029-9
viola, 1995, empirical entropy manipulation for real-world problems, Proc Neural Inform Processing Syst (NIPS 8) Conf, 851
linsker, 1988, toward an organizing principle for a layered perceptual network, Neural Information Processing Systems, 485
10.1162/neco.1997.9.7.1457
deco, 1996, An Information-Theoretic Approach to Neural Computing, 10.1007/978-1-4612-4016-7
renyi, 1970, Probability Theory
10.1016/S0925-2312(02)00526-X
csiszár csiszar, 1981, Information Theory Coding Theorems for Discrete Memoryless Systems
haykin, 1984, Introduction to Adaptive Filters
10.1109/NNSP.1998.710621
luenberger, 1973, Linear and Nonlinear Programming
10.1038/323533a0