A study for multi-objective fitness function for time series forecasting with intelligent techniques

Authors:
Aranildo Rodrigues Lima Junior
Affiliations:
Federal University of Pernambuco, Recife, Brazil
Venue:
Proceedings of the 10th annual conference companion on Genetic and evolutionary computation
Year:
2008

Citing 3
Cited 1

Genetic programming: on the programming of computers by means of natural selection

Genetic programming: on the programming of computers by means of natural selection
Introduction to Evolutionary Computing

Introduction to Evolutionary Computing
A new evolutionary method for time series forecasting

GECCO '05 Proceedings of the 7th annual conference on Genetic and evolutionary computation

Combining artificial neural network and particle swarm system for time series forecasting

IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks

Quantified Score

Hi-index	0.01

Visualization

Abstract

This paper presents an study of a Hybrid method for time series prediction, called GRASPES, based on Greedy Randomized Adaptive Search Procedure (GRASP) Algorithm and Evolutionary Strategies (ES) concepts for tuning of the structure and parameters of an Artificial Neural Network (ANN). An experimental investigation with two time series is conducted and the results achieved are discussed and compared to other works reported in the literature. Distinct fitness functions evaluations are shown, instead of conventional MSE or NMSE based fitness functions evaluation. This results shown that small changes of the fitness function evaluation could lead to a significantly improved performance.