Fuzzy adaptive Q-learning method with dynamic learning parameters

An active search in the reinforcement learning disturbs the learning process when learning proceeds and converges to a partial search area. Therefore, it is important to balance between searching behavior of the unknown knowledge and using behavior of the obtained knowledge. In this research, we propose an adaptive Q-learning method tuning learning parameters of the reinforcement learning by fuzzy rules. We also report some results of artificial ants simulation using this method.