A Brand-New Simple, Fast, and Effective Residual-Based Method for Radial Basis Function Neural Networks Training

The radial basis function (RBF) neural network is a type of universal approximator, and has been widely used in various fields. Improving the training speed compactness RBF networks are critical for promoting their applications. In present study, we propose simple, fast, effective method, which based on residual extreme points neighborhoods (thus called REN method short this paper). calculates centers widths through two-level iterative process, realizes two main functionalities, namely 1) adding multiple within one pass whole data set, 2) calculating specifically each center. use algorithm does not need any parameter adjustments, models approximation or classification can be obtained by only run. performance proposed compared with classic powerful orthogonal least squares (OLS) algorithm. By reaching same accuracies, trains 50 320 times faster, chirp (0˜50 Hz, 2 s, 1 kHz, 2001 samples) two-dimensional peaks (2401 signal tasks respectively, than OLS does, number reduced half. When incorporating centers, achieves accuracies up to 3 orders magnitude higher best results task real discrete breast cancer data, both methods result comparable many existent methods, but advantages fast speeds no requirements adjustments. study may potentially large scale applications that require high model performances.