Diagonal Log-Normal Generalized RBF Neural Network for Stock Price Prediction
Stock price prediction is one of the most important topics in financial engineering. In this paper, for stock closing price prediction, we propose a diagonal log-normal generalized RBF neural network in which the diagonal log-normal density functions serve as the RBFs. Specifically, it utilizes the dynamic split-and-merge EM algorithm to select the number of hidden units (or RBFs) as well as the initial values of the parameters, and implements a synchronous LMS learning algorithm for parameter learning. It is demonstrated by the experiments that the diagonal log-normal generalized RBF neural network has a competitive performance on stock closing price prediction.
KeywordsRBF neural network Log-normal distributions EM algorithm Stock price prediction
Unable to display preview. Download preview PDF.
- 1.Jiang, Y., Lin, Y.: The Application of RBF Neural Networks to Stock Price Forecasting. Mind and Computation 1(4), 413–419 (2007) (in Chinese)Google Scholar
- 2.Liu, H., Bai, Y.: Analysis of AR Model and Neural Network for Forecasting Stock Price. Mathmatics in Practice and Theory 41(4), 14–19 (2011) (in Chinese)Google Scholar
- 3.Liu, S., Ma, J.: The Application of Diagonal Generalized RBF Neural Network to Stock Price Prediction. China Sciencepaper Online (2014) (in Chinese)Google Scholar
- 4.Tsang, P.M., Kwok, P., Choy, S.O., et al.: Design and Implementation of NN5 for Hong Kong Stock Price Forcasting. Enginerring Applications of Artificial Intelligence 20(4), 453–461 (2007)Google Scholar
- 5.Fu, C., Fu, M., Que, J.: Prediction of Stock Price Base on Radial Basic Function Neural Networks. Technological Development of Enterprise 23(4), 14–15 (2004) (in Chinese)Google Scholar
- 6.Zheng, P., Ma, Y.: RBF Neural Network Based Sock Market Modeling and Forecasting. Journal of Tianjin University 33(4), 483–486 (2000) (in Chinese)Google Scholar
- 8.Broomhead, D.S., Lowe, D.: Mutivariable Functional Interpolation and Adaptive Networks. Complex Systems 2, 321–355 (1988)Google Scholar
- 10.Ma, J., Qing, C.: Diagonal Generalized RBF Neural Network and Nonelinear Time Series Prediction. Journal of Signal Processing 29(12), 1609–1614 (2013) (in Chinese)Google Scholar
- 11.Yu, Y.: The Application of the Lognormal Distribution in the Stock Price Models. Journal of Langfang Teachers College (Natural Science Edition) 12(5), 69–72 (2012) (in Chinese)Google Scholar
- 12.Long, S., Xiang, L.: Emirical Analysis of Stock Price’s Lognormal Distribution. Journal of Huanggang Normal University 33(3), 9–11 (2013) (in Chinese)Google Scholar
- 13.Wang, L., Ma, J.: A Kurtosis and Skewness Based Criterion for Model Selection on Gaussian Mixture. In: The 2nd International Conference on Biomedical Engineering and Information, pp. 1–5 (2009)Google Scholar
- 14.Wang, L., Ma, J.: Efficient Training of RBF Networks via the Kurtosis and Skewness Minimization Learning Alogrithm. Journal of Theoretical and Applied Information Technology 48(1), 496–504 (2013)Google Scholar