论文信息 - A RPCL-CLP architecture for financial time series forecasting

A RPCL-CLP architecture for financial time series forecasting

In this paper, we propose a new architecture based on the rival penalized competitive learning algorithm (RPCL) of Xu, Krzyzak and Oja (1993) and combined linear prediction method (CLP). The performance of RPCL-CLP is insensitive to the initial number of cluster nodes selected. Experimental results show that it is robust in long-term prediction for financial time series forecasting.

Yiu-ming Cheung | Lei Xu | Wai Man Leung

[1] M. F. Tenorio,et al. A ClusNet architecture for prediction , 1993, IEEE International Conference on Neural Networks.

[2] Erkki Oja,et al. Rival penalized competitive learning for clustering analysis, RBF net, and curve detection , 1993, IEEE Trans. Neural Networks.

[3] Stanley C. Ahalt,et al. Competitive learning algorithms for vector quantization , 1990, Neural Networks.