THE ESTIMATION OF PRIOR FROM FISHER INFORMATION
THE ESTIMATION OF PRIOR FROM FISHER INFORMATION作者机构:Department of Mathematics Lanzhou University Lanzhou China Department of Mathematics & Statistics University of Nebraska-Lincoln USA Basic Department People’s Police University of China Beijing
出 版 物:《Applied Mathematics(A Journal of Chinese Universities)》 (高校应用数学学报(英文版)(B辑))
年 卷 期:1994年第9卷第2期
页 面:109-120页
核心收录:
学科分类:02[经济学] 0202[经济学-应用经济学] 020208[经济学-统计学] 0701[理学-数学]
主 题:62F15 Divergence, entropy Fisher information prior
摘 要:In Bayesian analysis, the maximum entropy prior is commonly used. But in some cases, the maximum entropy prior does not exist or has unpleasant jumps, In this paper, Fisher-information is used to measure the divergence and to derive the unknown prior. To a certain degree, this measure overcomes the shortcomings of the maximum entropy. The Fisher prior aways exists and it is continuous under the general requirements.