作者: Lee Sukhan , Kil
关键词: Artificial neural network 、 Pattern recognition 、 Gaussian 、 Gradient descent 、 Feed forward 、 Function (mathematics) 、 Artificial intelligence 、 Computer science 、 Set (abstract data type) 、 Domain (mathematical analysis) 、 Pattern recognition (psychology) 、 Algorithm
摘要: The authors present a multilayer feedforward network, called the Gaussian potential function network (GPFN), performing association or classification based on set of potentially fields synthesized over domain input space by number units (GPFUs). A GPFU as basic component GPFN is designed to generate form field. weighted summation generated suitable GPFUs provides an arbitrary shape field space. also detailed learning algorithm for GPFN. Learning consists determination minimally necessary and adjustment locations shapes individual defined well weights. control effective radius GPFUs, while parameter gradient descent procedure. >