New Strategies for Initialization and Training of Radial Basis Function Neural Networks

In this paper we proposed two new strategies for initialization and training of Radial Basis Function (RBF) Neural Network. The first approach takes into consideration the "error" between the input vector p of the network and the x-axis, which are the centers of radial functions. The second approach takes into account the "error" between the input vector p and the network output. In order to check the performances of these strategies, we used Brazilian financial market data for the RBF networks training, specifically the adjusted prices of the 10 greater weighted shares in the Bovespa index at the time of data collection ‑ from April 8th, 2009 to October 31th, 2014. The first approach presented a 52% of improvement in the mean squared error (MSE) compared to the standard RBF network, while the improvement for the second approach was 38%. The strategies proved to be consistent for the time series tested, in addition to having a low computational cost. It is proposed that these strategies be improved by testing them with the Levenberg-Marquardt algorithm.

[1]  Ruimin Zhang,et al.  The application of dynamic intelligent neural network in time series forecasting , 2011, 2011 International Conference on Electrical and Control Engineering.

[2]  Shigeyuki Hamori.  An Empirical Investigation of Stock Markets: The CCF Approach , 2003 .

[3]  Teresa Bernarda Ludermir,et al.  Optimization of the weights and asymmetric activation function family of neural network for time series forecasting , 2013, Expert Syst. Appl..

[4]  Ignacio J. Turias,et al.  Hybrid approaches based on SARIMA and artificial neural networks for inspection time series forecasting , 2014 .

[5]  Nikolaos Kourentzes,et al.  Neural network ensemble operators for time series forecasting , 2014, Expert Syst. Appl..

[6]  Çagdas Hakan Aladag,et al.  Fuzzy time series forecasting with a novel hybrid approach combining fuzzy c-means and neural networks , 2013, Expert Syst. Appl..

[7]  Çagdas Hakan Aladag,et al.  Fuzzy time series forecasting method based on Gustafson-Kessel fuzzy clustering , 2011, Expert Syst. Appl..

[8]  Yaochu Jin,et al.  Evolutionary multi-objective generation of recurrent neural network ensembles for time series prediction , 2014, Neurocomputing.

[9]  Hujun Yin,et al.  Time series prediction with a non-causal neural network , 2014, 2014 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr).

[10]  Li Song,et al.  Prediction for chaotic time series of optimized BP neural network based on modified PSO , 2014, The 26th Chinese Control and Decision Conference (2014 CCDC).

[11]  R. K. Agrawal,et al.  PSO based Neural Networks vs. traditional statistical models for seasonal time series forecasting , 2013, 2013 3rd IEEE International Advance Computing Conference (IACC).

[12]  Friedhelm Schwenker,et al.  Three learning phases for radial-basis-function networks , 2001, Neural Networks.

[13]  Rohitash Chandra,et al.  Multi-objective cooperative coevolution of neural networks for time series prediction , 2014, 2014 International Joint Conference on Neural Networks (IJCNN).

[14]  Michal Tkác,et al.  Artificial neural networks in business: Two decades of research , 2016, Appl. Soft Comput..

[15]  Diyar Akay,et al.  Comparison of direct and iterative artificial neural network forecast approaches in multi-periodic time series forecasting , 2009, Expert Syst. Appl..

[16]  Yusuf Leblebici,et al.  Review of advances in neural networks: Neural design technology stack , 2016, Neurocomputing.

[17]  Fan Yu,et al.  Time Delay Prediction Method Based on EMD and Elman Neural Network , 2014, 2014 Sixth International Conference on Intelligent Human-Machine Systems and Cybernetics.

[18]  Wei Du,et al.  Time series forecasting by neural networks: A knee point-based multiobjective evolutionary algorithm approach , 2014, Expert Syst. Appl..

[19]  Mark Beale,et al.  Neural Network Toolbox™ User's Guide , 2015 .

[20]  Xiaodong Li,et al.  Time series forecasting by evolving artificial neural networks with genetic algorithms, differential evolution and estimation of distribution algorithm , 2011, Neural Computing and Applications.

[21]  Tao Chen,et al.  Back propagation neural network with adaptive differential evolution algorithm for time series forecasting , 2015, Expert Syst. Appl..

[22]  Paulo Cortez,et al.  Evolutionary optimization of sparsely connected and time-lagged neural networks for time series forecasting , 2014, Appl. Soft Comput..

[23]  Svitlana Galeshchuk,et al.  Neural networks performance in exchange rate prediction , 2016, Neurocomputing.