摘要
基于贝叶斯准则对二元假设检验问题进行研究,在附加先验概率条件下,提出了最小二乘意义下的最佳先验概率。将该方法与极大极小准则对比分析,给出了任意先验概率条件下的最佳先验概率选取方法,并进行仿真验证。仿真结果表明,根据最佳先验概率进行判决,可以在先验概率未知时有效降低判决代价。
The problem of the binary hypothesis test is studied based on the Bayesian Criterion and the optimal prior probability of least square is proposed in additional prior probability conditions. By comparing and analyzing the method and Maxmini criterion,an approach to determining the best prior probability under different prior probabilities is given and verified by simulation. The simulation results show that the total average cost of the judgment is effectively reduced using the best prior probability when the prior probability is unknown.
引文
[1] HOBELLAH I Y,VARSHNCY P K.Distributed bayesiansignal detection[J].TEEE Trans.on Information Theory,1989,35(5):995-1000.
[2]钱国栋,张振华,刘文松.基于贝叶斯准则的集中式与分布式检测对比研究[J].计算机与数字工程,2015,43(2):215-219.
[3]刘苏君,杨婷,唐明春,等.基于贝叶斯准则的随机共振算法研究[J].电子与信息学报,2017,39(2):293-299.
[4]吉彦军,杜兴民.基于MATLAB的最小平均代价(MAC)研究[J].弹箭与制导学报,2004,24(2):215-217.
[5]王国宏,毛士艺,何友,等.先验概率和代价函数均模糊时基于贝叶斯最小风险准则的分布式决策融合[J].电子学报,1999,27(12):35-38.
[6]张明友.信号检测与估计[M].北京:电子工业出版社,2011.