Artificial Life Environment Modeled by Dynamic Fuzzy Cognitive Maps

This paper presents an artificial life environment based on dynamic fuzzy cognitive maps (DFCMs) and inspired by multiagent systems, machine learning, and concepts from classical fuzzy cognitive map theory. The proposed architecture includes features such as a reinforcement learning algorithm to dynamically fine-tune the weights of the DFCM, a finite states machine, governing the behavior of the creatures by adding/removing concepts into the DFCM, and others. These features are used to add adaptability to the artificial creatures (agents) in a simulated hunter-prey environment with synthetic data. Some experiments carried out in a simulated virtual environment have shown promising results for further research in the subject of this paper.