Book Review: Regularized Radial Basis Function Networks: Theory and Applications

Problem Description -- Review of Current Approaches -- Traditional ANN Approaches -- Limitations of the ANN Approaches -- Kernel Regression Approach -- Limitations of the Kernel Regression Approach -- Guided Tour of Book -- Basic Tools -- Asymptotic Equivalence of NWRE to Regularized Strict Interpolation RBFN via Constructive Approximation -- Relationship between R[subscript 2] and R[subscript 2] -- Implications for Regularized Strict Interpolation RBFN Estimation -- Asymptotic Optimality with Respect to Mean-Squared Error -- Consistency with Respect to Mean-Squared Error over Compacta -- Probability Estimation and Pattern Classification -- Problem Description -- Review of Current Approaches -- Traditional ANN Approaches -- Kernel-Based Approaches -- Theoretical Results for Regularized Probability Estimates -- Consistency of Probability Estimates in Mean-Square and Bayes Risk -- Nonlinear Time-Series Prediction -- Problem Description -- Review of Current Approaches -- Traditional ANN Approaches -- Kernel Regression Approach -- Consistency of Prediction -- Recursive Updating for Regularized RBFN Predictors -- Augmented (Infinite Memory) Case -- Fixed-Size (Finite Memory) Case -- Application to Speech Prediction -- Experiment 1: Partial Update Algorithm for Fixed-Size Networks -- Experiment 2: Full Update Algorithm for Fixed-Size Networks -- Nonlinear State Estimation -- Problem Description -- ANN Approach -- Review of Current Approaches -- Proposed Approach -- Experimental Results -- Comparison to the SDE Approach.