作者: Qi Wu , Rob Law , Edmond Wu , Jinxing Lin
DOI: 10.1016/J.INS.2013.02.017
关键词:
摘要: In this paper, the relationship between Gaussian noise and loss function of support vector regression machine (SVRM) is analyzed, then a proposed to reduce effect such on estimates. Since @e-insensitive cannot noise, novel machine, g-SVRM, proposed, chaotic particle swarm optimization (CPSO) algorithm developed estimate its unknown parameters. Finally, hybrid-forecasting model combining g-SVRM with CPSO forecast multi-dimensional time series. The results two experiments demonstrate feasibility approach.