Singular Value Shrinkage Priors for Bayesian Prediction
Authors:
- Takeru MatsudaTMTakeru Matsuda
- Fumiyasu KomakiFKFumiyasu Komaki
Preprint
Abstract
We develop singular value shrinkage priors for the mean matrix parameters in the matrix-variate normal model with known covariance matrices. Our priors are superharmonic and put more weight on matrices with smaller singular values. They are a natural generalization of the Stein prior. Bayes estimators and Bayesian predictive densities based on our priors are minimax and dominate those based on the uniform prior in finite samples. In particular, our priors work well when the true value of the parameter has low rank.Keywords
All Related Versions
This publication has 7 references indexed in Scilit:
- Extended Matrix Variate Hypergeometric Functions and Matrix Variate DistributionsInternational Journal of Mathematics and Mathematical Sciences, 2015
- Admissibility and minimaxity of Bayes estimators for a normal mean matrixJournal of Multivariate Analysis, 2008
- Bayesian shrinkage prediction for the regression problemJournal of Multivariate Analysis, 2008
- PREDICTIVE DENSITY ESTIMATION FOR MULTIPLE REGRESSIONEconometric Theory, 2008
- Improved prediction for a multivariate normal distribution with unknown mean and varianceAnnals of the Institute of Statistical Mathematics, 2008
- Improved minimax predictive densities under Kullback–Leibler lossThe Annals of Statistics, 2006
- A shrinkage predictive distribution for multivariate Normal observablesBiometrika, 2001