Research On The Multi-agent Learning Model Based On DFL

Agent learning is one of the hottest proboles in recent years. This thesis presented a multi-agent learning model based on Dynamic Fuzzy Logic (DFL) because of the dynamic fuzzy characteristics of the agent's properties and the environment during the process of the agent learning. This model included the agent intelligence model, the agent intelligence states and their axioms, the agent working theories, single-agent learning algrithms, multi-agent learning algrithms, and so on. It also applied these algrithms to the actual problems and achieved the cards game based on DFL.Therefore, the main characteristics of this thesis are:(1) Advance an agent intelligence model based on DFL and this model is the base of researching agent learning;(2) Advance the single-agent learning algorithms and the multi-agent learning algorithms based on DFL, enrich and develop the basic content of agent learning further;(3) Give an example system: a cards game. Of course, all the work is tentive and much of them need advanced research. For example, the learning algorithms can be improved, and we need apply the multi-agent learning model to more fields, and so on.