Time series prediction method based on LS-SVR with modified gaussian RBF

Authors:
Yangming Guo;Xiaolei Li;Guanghan Bai;Jiezhong Ma
Affiliations:
School of Computer Science and Technology, Northwestern Polytechnical University, Xi'an, China;School of Computer Science and Technology, Northwestern Polytechnical University, Xi'an, China;Department of Mechanical Engineering, University of Alberta, Edmonton, Canada;School of Computer Science and Technology, Northwestern Polytechnical University, Xi'an, China
Venue:
ICONIP'12 Proceedings of the 19th international conference on Neural Information Processing - Volume Part II
Year:
2012

Citing 3
Cited 0

Predicting Time Series with Support Vector Machines

ICANN '97 Proceedings of the 7th International Conference on Artificial Neural Networks
KCS-new kernel family with compact support in scale space: formulation and impact

IEEE Transactions on Image Processing
Trend Time–Series Modeling and Forecasting With Neural Networks

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

LS-SVR is widely used in time series prediction. For LS-SVR, the selection of appropriate kernel function is a key issue, which has a great impact with the prediction accuracy. Compared with some other feasible kernel functions, Gaussian RBF is always selected as kernel function due to its good features. As a distance functions-based kernel function, Gaussian RBF also has some drawbacks. In this paper, we modified the standard Gaussian RBF to satisfy the two requirements of distance functions-based kernel functions which are fast damping at the place adjacent to the test point and keeping a moderate damping at infinity. The simulation results indicate preliminarily that the modified Gaussian RBF has better performance and can improve the prediction accuracy with LS-SVR.