摘要
为提高网络认知的准确度,采用双层贝叶斯网络模型对网络参数进行层次化描述;采用强化学习推理算法对模型的条件概率表进行分级和学习,删除冗余信息,更准确地反映网络参数间的依赖关系,保证网络认知算法的准确度。经仿真分析,证明算法能够更好地描述网络参数的依赖信息,具有较高的推理准确度。
In order to effectively improve the accuracy of cogntive network, this paper used a double-layer Bayesian networks model to describe the relations among different variables through a certain hierarchy. Thus, inference would become more accu- rate. It used reinforcement learning algorithm to learn and classify conditional probability table in the process of inference of these variables. This way could improve the inference accuracy and deleted redundant information, and it could more accurately reflect the dependent relationships among network parameters, guaranteed more conducive to precise inference. The simulation analysis proves the effectiveness of the algorithm.
出处
《计算机应用研究》
CSCD
北大核心
2014年第5期1320-1323,共4页
Application Research of Computers
基金
国家"863"计划资助项目(2011AA01A109)
关键词
认知网络
贝叶斯网络
分层模型
增强学习
cognitive network
Bayesian network
double-layer
reinforcement learning
