Title :
Model comparisons and predictive mean computations for hierarchical Bayesian neural nets: quadratic approximation vs. MCMC
Author :
Nakajima, Y. ; Asano, M. ; Nakada, Y. ; Matsumoto, T.
Author_Institution :
Waseda Univ., Tokyo, Japan
Abstract :
The article is a first step toward an attempt to demonstrate the validity of quadratic approximations (QAP) of computing marginal likelihood as well as predictive distributions for the hierarchical Bayesian scheme by using MCMC (Markov chains Monte Carlo). At least for the simple examples considered, the QAP gives reasonable results for marginal likelihood and predictive distributions. More elucidation is necessary to further study the issues for more complicated problems including nonlinear time series prediction problems
Keywords :
Bayes methods; Markov processes; Monte Carlo methods; approximation theory; neural nets; MCMC; Markov chains Monte Carlo; QAP; hierarchical Bayesian neural nets; hierarchical Bayesian scheme; marginal likelihood; model comparisons; nonlinear time series prediction problems; predictive distributions; predictive mean computations; quadratic approximation; Annealing; Approximation algorithms; Bayesian methods; Context modeling; Distributed computing; Feedforward neural networks; Monte Carlo methods; Neural networks; Prediction algorithms; Predictive models;
Conference_Titel :
Neural Information Processing, 1999. Proceedings. ICONIP '99. 6th International Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-7803-5871-6
DOI :
10.1109/ICONIP.1999.843975
