Two-stage learning algorithm for fuzzy cognitive maps

Abstract

A two-stage learning algorithm based on Hebbian learning rule and evolutionary computation technique is presented in this paper for training Fuzzy Cognitive Maps. Fuzzy Cognitive Maps is a soft computing technique for modeling complex systems, which combines the synergistic theories of neural networks and fuzzy logic. The methodology of developing Fuzzy Cognitive Maps (FCMs) relies on human expert experience and knowledge, but still exhibits weaknesses in utilization of learning methods. We investigate in this work a coupling of Differential Evolution algorithm and Unsupervised Hebbian learning algorithm, using both the global search capabilities of Evolutionary techniques and the effectiveness of the Nonlinear Hebbian learning rule. The proposed algorithm applied successfully in a real-world process control problem. Experimental results suggest that the two-stage learning strategy is capable to train FCMs effectively leading the system to desired steady states and determining the appropriate weight matrix.