Selecting artificial neural network inputs using particle swarm optimization

Lawrence Allen Bliss, Pace University

Abstract

Much work has been clone in the area of configuring Artificial Neural Network (ANN) topology automatically using soft computing techniques such as Genetic Algorithms (GA). However, little time has been spent by researches on selecting the proper inputs to the ANN. Neural Networks used to predict the behavior of Dynamical Systems often have a choice of input information, much of which is redundant. Selecting a minimal set of inputs that produce acceptable behavior results in a lower cost, solution. Researchers using trial and error methods currently do this input selection manually. The work presented in this paper shows several methods of automatically selecting a small set of inputs from a large candidate population. Particle Swarm Optimization is the primary network optimization technique used for processing the ANN configuration. ^

Subject Area

Physics, Fluid and Plasma|Computer Science

Recommended Citation

Lawrence Allen Bliss, "Selecting artificial neural network inputs using particle swarm optimization" (January 1, 2003). ETD Collection for Pace University. Paper AAI3106822.
http://digitalcommons.pace.edu/dissertations/AAI3106822

Share

COinS

Remote User: Click Here to Login (must have Pace University remote login ID and password. Once logged in, click on the View More link above)