Model choice in generalised linear models: a Bayesian approach via Kullback-Leibler projections

A Kullback-Leibler Divergence for Bayesian Model Diagnostics

“… This paper considers a Kullback-Leibler distance (KLD) which is asymptotically equivalent to the KLD by Goutis and Robert [1] when the reference model (in comparison to a competing fitted model) is correctly specified and that certain regularity conditions hold true (ref. Akaike [2]).…”

Current Advances in Uncertainty Estimation of Earth Observation Products of Water Quality

“…The use of flat or improper priors, e.g. uniform distribution, may invalidate the derivation of the posterior probability (Goutis & Robert, 1998). According to the maximum entropy principle (Jaynes, 1957a; a proper prior probability distribution should have the maximum entropy provided by the IOP-triplet.…”

Compatible priors for Bayesian model comparison with an application to the Hardy–Weinberg equilibrium model

“…This prior was considered by McCulloch and Rossi (1992). Goutis and Robert (1998) and Dupuis and Robert (2003) use the notion of a KL-projection to perform Bayesian hypothesis testing (or model selection), although they do not resort to KL-projection priors.…”

“…This idea has proved to be especially fruitful in the framework of model choice/comparison. For instance, Goutis and Robert (1998) and Bernardo and Rueda (2002) use the expectation, relative to the posterior distribution of θ , of a measure of divergence between a model and a submodel in order to assess the validity of model simplification. Specifically, the latter paper uses a decision-theoretic approach to model choice, based on the concept of intrinsic discrepancy, and the corresponding reference prior, which only depends on the structure of the model.…”