Improving the performance of autonomous agents processed by modified fuzzy cognitive maps
What is it about?
The paper introduces using decision-making method for autonomous agents processed by matrix-based method. This improves the performance of the model while keeping almost the same quality of decision-making. It also allows learning through evolution. In this paper, experiments are done on artificial life model, but it can be used for any autonomous system.
Why is it important?
Nowadays, research of artificial intelligence aims for maximum complexity with neural networks, which leads to black box solutions nobody can read or understand. On the other hand, rest of this field are mostly hardcoded simple rules which doesn't bring good adaptability or learning. In the middle, there are matrix-based solutions like fuzzy cognitive maps modified for autonomous agents (AA-FCM). This method is human-readable, able to learn and very strong in decision-making. This paper focuses on the computational complexity of this method and improves it by combination with analytic hierarchy process (AHP).