Emotional Temporal Difference Learning Based Multi-layer Perceptron Neural Network Application to a Prediction of Solar Activity

摘要

Nonlinear time series prediction has in recent years, been the subjects of many methodological and applied studies in the fields of system identification and nonlinear prediction. An important benchmark has been the prediction of solar activity with the markup increase in the practical importance of space weather forecasting; its motivation has risen far beyond more methodological concerns. In this paper, we have used a bounded rationality decision-making procedure, whose utility has been demonstrated in several identification and control tasks, for predicting sunspot numbers. An emotional temporal difference learning based multi layer perceptron neural network is introduced and applied to the prediction task.