Accession Number : ADA290862

Title :   Neural Network Construction Using Evolutionary Search.

Descriptive Note : Professional paper,

Corporate Author : NAVAL COMMAND CONTROL AND OCEAN SURVEILLANCE CENTER RDT AND E DIV SAN DIEGO CA

Personal Author(s) : McDonnell, J. R. ; Page, W. C. ; Waagen, D. E.

PDF Url : ADA290862

Report Date : DEC 1994

Pagination or Media Count : 10

Abstract : This work investigates the application of evolutionary search for training candidate hidden units in cascade-correlation learning architectures. A hybrid evolutionary search algorithm which implements techniques from evolutionary programming and evolution strategies is proposed. This approach is evaluated on selected low-dimensional examples which are non-linearly separable. (AN)

Descriptors :   *NEURAL NETS, ALGORITHMS, EXPERIMENTAL DATA, OPTIMIZATION, DISTRIBUTED DATA PROCESSING, COMPUTER PROGRAMMING, COMPUTER ARCHITECTURE, LEARNING MACHINES, CORRELATION, SEARCHING, HYBRID SYSTEMS, PATTERN RECOGNITION, ARTIFICIAL INTELLIGENCE, KNOWLEDGE BASED SYSTEMS, DISCRIMINATE ANALYSIS.

Subject Categories : Cybernetics

Distribution Statement : APPROVED FOR PUBLIC RELEASE