Title :
Neural network tools for stellar light prediction
Author :
Cholewo, Tomasz J. ; Zurada, Jacek M.
Author_Institution :
Dept. of Electr. Eng., Louisville Univ., KY, USA
Abstract :
This paper presents a comparative study of state-of-the-art neurocomputing methods applied to several benchmark time series, including the white dwarf light curve. The goal is to determine which of the predictive models work best for data from natural sources. The emphasis is on using a unified methodology for selection of the best architectures among those used for comparison. The specific architectures considered are a Finite Impulse Response (FIR) network and three types of layered recurrent networks: Jordan, Elman, and extended Elman. An enhancement of a FIR network allowing selection of weights with relevant time delays only is also presented. Our approach is applied to two benchmark prediction problems: the Wolfer sunspot number data and a white dwarf light curve. Results show that the best predictions are obtained using a FIR neural network
Keywords :
FIR filters; astronomical techniques; astronomy computing; autoregressive moving average processes; backpropagation; conjugate gradient methods; multilayer perceptrons; neural net architecture; recurrent neural nets; stellar photometry; stellar radiation; sunspots; time series; white dwarfs; ARMA models; Elman network; Jordan network; Wolfer sunspot number data; batch mode adaptation; benchmark time series; data from natural sources; extended Elman network; finite impulse response network; inverse pruning; layered recurrent networks; multilayer perceptron; network size selection; neural network tools; neurocomputing methods; nonlinear models; predictive models; scaled conjugate gradient method; selection of weights; sequential network; stellar light prediction; temporal backpropagation; unified methodology; white dwarf light curve; Cost function; Finite impulse response filter; Guidelines; In vitro fertilization; Learning systems; Multi-layer neural network; Neural networks; Neurons; Nonlinear filters; Predictive models;
Conference_Titel :
Aerospace Conference, 1997. Proceedings., IEEE
Conference_Location :
Snowmass at Aspen, CO
Print_ISBN :
0-7803-3741-7
DOI :
10.1109/AERO.1997.574895
