Active Hebbian learning algorithm to train fuzzy cognitive maps

Abstract Fuzzy cognitive map is a soft computing technique for modeling systems, which combines synergistically the theories of neural networks and fuzzy logic. Developing of fuzzy cognitive map (FCM) relies on human experience and knowledge, but still exhibits weaknesses in utilization of learning methods. The critical dependence on experts and the potential uncontrollable convergence to undesired steady-states are important deficiencies to manage FCMs. Overcoming these deficiencies will improve the efficiency and robustness of the FCM methodology. Learning and convergence algorithms constitute the mean to improve these characteristics of FCMs, by modifying the values of cause–effect weights among concepts. In this paper a new learning algorithm that alleviates the problem of the potential convergence to a steady-state, named Active Hebbian Learning (AHL) is presented, validated and implemented. This proposed learning procedure is a promising approach for exploiting experts' involvement with their subjective reasoning and at the same time improving the effectiveness of the FCM operation mode and thus it broadens the applicability of FCMs modeling for complex systems.