Research topic: Computer-supported assessments of biological systems including aspects such as: adaptation (evolution, development, life-long learning), behavior, and cognition
Research methods: Computational intelligence methods such as evolutionary computation, artificial neural-networks, fuzzy logic, and their hybridizations.
Main Projects in the lab include:
- Neuro-fuzzy Inferencing about Natural Systems – The development of a novel adaptive-neuro-fuzzy inference method for understanding the behavior of biological systems. In collaboration with Prof. Amir Ayali, we have already shown the effectiveness of this generic approach for the case of a marching locust in a swarm.
- Multi-payoff Games – In the past utility-based game theory has been used for the understanding of natural systems. In collaboration with Dr. G. Avigad and others, we have developed a non-utility based approach to multi-payoff games (games involving conflicting objectives for each player). We postulate that our recent achievements in defining rationalizable strategies in such games may lead to some new developments in understanding natural systems.
- Computational Neuroevolution – We have developed several unique approaches to artificially evolve neural-networks. For example, we suggested a unique algorithm for multi-objective topology and weight evolution of recurrent neural networks. While developed for robotics, we suggest that it may be used for understanding natural systems. Note: The postulations in item 2 and 3 are in accordance with our theory of multi- competence cybernetics. Initial presentation of this theory can be found in: Moshaiov, A. “Multi-competence Cybernetics: The Study of Multi-objective Artificial Systems and Multi- fitness Natural Systems.” In Multiobjective Problem Solving from Nature. Springer Berlin Heidelberg, pp. 285-304, 2008.