Brain Emotional Learning Networks with Applications

Abstract

Emotion plays a vital role in human learning, memory and decision-making, which have attracted wide attention in various research fields. To improve the learning speed and performance, a novel Brain Emotional Learning Network (BELN) is developed in this paper, which adds the affective neurons into the model of brain emotional learning and ameliorates the structure of the model. The newly proposed BELN enjoys lower computational complexity, in the sense that two affective coefficients named “anxiety” and “confidence” are added to simulate the changes of emotion in human learning process, which improves the online learning speed of the network. In order to verify the effectiveness of the network, a numerical example and the predictive control of petroleum heating process in a co-current tubular heat exchanger have been studied. The results show that compared with the BP neural network, BELN has better learning speed and recognition ability.