Current browse context:
stat.ME
Change to browse by:
References & Citations
Statistics > Methodology
Title: Loss-Based Variational Bayes Prediction
(Submitted on 29 Apr 2021 (v1), last revised 12 May 2022 (this version, v2))
Abstract: We propose a new approach to Bayesian prediction that caters for models with a large number of parameters and is robust to model misspecification. Given a class of high-dimensional (but parametric) predictive models, this new approach constructs a posterior predictive using a variational approximation to a generalized posterior that is directly focused on predictive accuracy. The theoretical behavior of the new prediction approach is analyzed and a form of optimality demonstrated. Applications to both simulated and empirical data using high-dimensional Bayesian neural network and autoregressive mixture models demonstrate that the approach provides more accurate results than various alternatives, including misspecified likelihood-based predictions.
Submission history
From: David Frazier [view email][v1] Thu, 29 Apr 2021 00:36:08 GMT (45kb,D)
[v2] Thu, 12 May 2022 11:31:30 GMT (52kb)
Link back to: arXiv, form interface, contact.