Abstrakt

A method of agent running mechanism based on bayesian network

Yu Lin, Zhang Jie, Cai Jin

Bayesian Network has provided a convenient frame structure to express causal relationship, by regarding influence diagramas a special Bayesian Network, then the value of each decision variable is imposed externally to meet the goals, rather than derive fromdetermination probability of father node. Using influence diagram to describe the node types of directed acyclic graph has been studied, which includes Decision nodes, Chance nodes andValue node.Agent itself is also an environmentalmodel, variable (V E) is the belief of node in environment, utility node (UE) denotes the self-preference of Agent, which is the objective defined by multi-Agent self-organizing system.