基于进化HMM 模型的动态Agent 系统建模方法

提出一种基于隐式Markov模型（HMM）的进化建模方法．使用进化算法随机搜索HMM的模型空间，自动选择HMM的结构和参数-完成对动态智能体系统行为的建模，学习智能体对周围环境的分割和反映方式．实验结果表明，该方法可以很好地搜索HMM的模型空间，并且避免了人工确定HMM模型结构的困难和手工设计模型所需的多次反复．

Evolutionary Hidden Markov Modeling for Dynamic Agent Systems

An evolutionary approach is proposed to model dynamic agent systems using Hidden Markov Models(HMMs).An enhanced genetic algorithm is used to automatically learn the structure and parameters of the HMM,and the final HMM can represent the agent's behaviors by segmenting its environment with an appropriate manner.Experiments indicate that the new method is good at searching the global model parameter space of HMMs,it outperforms conventional optimal HMM topology design methods which has a tendency to stagnate on local optima and usually requires a priori knowledge from a field expert.