A recursive Renyi's entropy estimator

D. Erdogmus1, J.C. Principe1, Sung-Phil Kim1, J.C. Sanchez2
1Electrical & Computer Engineering Department, Computational Neuro-Engineering Laboratory, Gainesville, FL, USA
2Biomedical Engineering Department, University of Florida, Gainesville, FL, USA

Tóm tắt

Estimating the entropy of a sample set is required, in solving numerous learning scenarios involving information theoretic optimization criteria. A number of entropy estimators are available in the literature; however, these require a batch of samples to operate on in order to yield an estimate. We derive a recursive formula to estimate Renyi's (1970) quadratic entropy on-line, using each new sample to update the entropy estimate to obtain more accurate results in stationary situations or to track the changing entropy of a signal in nonstationary situations.

Từ khóa

#Entropy #Recursive estimation #Information theory #Adaptive systems #Biomedical computing #Biomedical engineering #Yield estimation #Digital communication #Neural networks #Stochastic processes

Tài liệu tham khảo

10.1109/NNSP.2001.943148 erdogmus, 2001, An On-Line Adaptation A lgorithm for Adaptive System Training with Minimum Error Entropy: Stochastic Information Gradient, Proc of Indenendent Component Analysis 2001 (LCA'O I) rcnyi, 1970 parzen, 1967, On Estimation of a Probabi Density Functio and Mode, Series Analysis Papers principe, 2000, Information Theoretic Learning, Unsupervised Adaptive Fjltering, 1 erdogmus, 2002, Information Theoretic Learning: Renyi's Entropy and Its Applications to Adapt, System Trajning 10.1109/TNN.2002.1031936 10.1109/TSP.2002.1011217 10.1088/0954-898X/3/2/009 fisher, 1997, NonlinearExtensjons to the Mjnimnm Ayerage Correlation Energy Filter linsker, 1988, An Application of the Principle of Maximum lnformation Preservation to Linear Systems 10.1162/neco.1993.5.1.45 10.1109/9.587329 10.1002/j.1538-7305.1948.tb01338.x erdogmus, 2001, Convergence Analysis of he Information Potential Criterion in Adaline Training, Proc Neural Networks for Signal Processing (NNSP), 123