likelihood-free inference by ratio estimation

“This approach for posterior estimation with generative models mirrors the approach of Gutmann and Hyvärinen (2012) for the estimation of unnormalised models. The main difference is that here we classify between two simulated data sets while Gutmann and Hyvärinen (2012) classified between the observed data and simulated reference data.”

A 2018 arXiv posting by Owen Thomas et al. (including my colleague at Warwick, Rito Dutta, CoI warning!) about estimating the likelihood (and the posterior) when it is intractable. Likelihood-free but not ABC, since the ratio likelihood to marginal is estimated in a non- or semi-parametric (and biased) way. Following Geyer’s 1994 fabulous estimate of an unknown normalising constant via logistic regression, the current paper which I read in preparation for my discussion in the ABC optimal design in Salzburg uses probabilistic classification and an exponential family representation of the ratio. Opposing data from the density and data from the marginal, assuming both can be readily produced. The logistic regression minimizing the asymptotic classification error is the logistic transform of the log-ratio. For a finite (double) sample, this minimization thus leads to an empirical version of the ratio. Or to a smooth version if the log-ratio is represented as a convex combination of summary statistics, turning the approximation into an exponential family,  which is a clever way to buckle the buckle towards ABC notions. And synthetic likelihood. Although with a difference in estimating the exponential family parameters β(θ) by minimizing the classification error, parameters that are indeed conditional on the parameter θ. Actually the paper introduces a further penalisation or regularisation term on those parameters β(θ), which could have been processed by Bayesian Lasso instead. This step is essentially dirving the selection of the summaries, except that it is for each value of the parameter θ, at the expense of a X-validation step. This is quite an original approach, as far as I can tell, but I wonder at the link with more standard density estimation methods, in particular in terms of the precision of the resulting estimate (and the speed of convergence with the sample size, if convergence there is).

ISBA 2016 [#5]

On Thursday, I started the day by a rather masochist run to the nearby hills, not only because of the very hour but also because, by following rabbit trails that were not intended for my size, I ended up being scratched by thorns and bramble all over!, but also with neat views of the coast around Pula.  From there, it was all downhill [joke]. The first morning talk I attended was by Paul Fearnhead and about efficient change point estimation (which is an NP hard problem or close to). The method relies on dynamic programming [which reminded me of one of my earliest Pascal codes about optimising a dam debit]. From my spectator’s perspective, I wonder[ed] at easier models, from Lasso optimisation to spline modelling followed by testing equality between bits. Later that morning, James Scott delivered the first Bayarri Lecture, created in honour of our friend Susie who passed away between the previous ISBA meeting and this one. James gave an impressive coverage of regularisation through three complex models, with the [hopefully not degraded by my translation] message that we should [as Bayesians] focus on important parts of those models and use non-Bayesian tools like regularisation. I can understand the practical constraints for doing so, but optimisation leads us away from a Bayesian handling of inference problems, by removing the ascertainment of uncertainty…

Later in the afternoon, I took part in the Bayesian foundations session, discussing the shortcomings of the Bayes factor and suggesting the use of mixtures instead. With rebuttals from [friends in] the audience!

This session also included a talk by Victor Peña and Jim Berger analysing and answering the recent criticisms of the Likelihood principle. I am not sure this answer will convince the critics, but I won’t comment further as I now see the debate as resulting from a vague notion of inference in Birnbaum‘s expression of the principle. Jan Hannig gave another foundation talk introducing fiducial distributions (a.k.a., Fisher’s Bayesian mimicry) but failing to provide a foundational argument for replacing Bayesian modelling. (Obviously, I am definitely prejudiced in this regard.)

The last session of the day was sponsored by BayesComp and saw talks by Natesh Pillai, Pierre Jacob, and Eric Xing. Natesh talked about his paper on accelerated MCMC recently published in JASA. Which surprisingly did not get discussed here, but would definitely deserve to be! As hopefully corrected within a few days, when I recoved from conference burnout!!! Pierre Jacob presented a work we are currently completing with Chris Holmes and Lawrence Murray on modularisation, inspired from the cut problem (as exposed by Plummer at MCMski IV in Chamonix). And Eric Xing spoke about embarrassingly parallel solutions, discussed a while ago here.

variational Bayes for variable selection

Xichen Huang, Jin Wang and Feng Liang have recently arXived a paper where they rely on variational Bayes in conjunction with a spike-and-slab prior modelling. This actually stems from an earlier paper by Carbonetto and Stephens (2012), the difference being in the implementation of the method, which is less Gibbs-like for the current paper. The approach is not fully Bayesian in that, not only an approximate (variational) representation is used for the parameters of interest (regression coefficient and presence-absence indicators) but also the nuisance parameters are replaced with MAPs. The variational approximation on the regression parameters is an independent product of spike-and-slab distributions. The authors show the approximate approach is consistent in both frequentist and Bayesian terms (under identifiability assumptions). The method is undoubtedly faster than MCMC since it shares many features with EM but I still wonder at the Bayesian interpretability of the outcome, which writes out as a product of estimated spike-and-slab mixtures. First, the weights in the mixtures are estimated by EM, hence fixed. Second, the fact that the variational approximation is a product is confusing in that the posterior distribution on the regression coefficients is unlikely to produce posterior independence.

O-Bayes15 [day #1]

So here we are back together to talk about objective Bayes methods, and in the City of Valencià as well.! A move back to a city where the 1998 O’Bayes took place. In contrast with my introductory tutorial, the morning tutorials by Luis Pericchi and Judith Rousseau were investigating fairly technical and advanced, Judith looking at the tools used in the frequentist (Bernstein-von Mises) analysis of priors, with forays in empirical Bayes, giving insights into a wide range of recent papers in the field. And Luis covering works on Bayesian robustness in the sense of resisting to over-influential observations. Following works of him and of Tony O’Hagan and coauthors. Which means characterising tails of prior versus sampling distribution to allow for the posterior reverting to the prior in case of over-influential datapoints. Funny enough, after a great opening by Carmen and Ed remembering Susie, Chris Holmes also covered Bayesian robust analysis. More in the sense of incompletely or mis-  specified models. (On the side, rekindling one comment by Susie and the need to embed robust Bayesian analysis within decision theory.) Which was also much Chris’ point, in line with the recent Watson and Holmes’ paper. Dan Simpson in his usual kick-the-anthill-real-hard-and-set-fire-to-it discussion pointed out the possible discrepancy between objective and robust Bayesian analysis. (With lines like “modern statistics has proven disruptive to objective Bayes”.) Which is not that obvious because the robust approach simply reincorporates the decision theory within the objective framework. (Dan also concluded with the comic strip below, whose message can be interpreted in many ways…! Or not.)

The second talk of the afternoon was given by Veronika Ročková on a novel type of spike-and-slab prior to handle sparse regression, bringing an alternative to the standard Lasso. The prior is a mixture of two Laplace priors whose scales are constrained in connection with the actual number of non-zero coefficients. I had not heard of this approach before (although Veronika and Ed have an earlier paper on a spike-and-slab prior to handle multicolinearity that Veronika presented in Boston last year) and I was quite impressed by the combination of minimax properties and practical determination of the scales. As well as by the performances of this spike-and-slab Lasso. I am looking forward the incoming paper!

The day ended most nicely in the botanical gardens of the University of Valencià, with an outdoor reception surrounded by palm trees and parakeet cries…

projective covariate selection

While I was in Warwick, Dan Simpson [newly arrived from Norway on a postdoc position] mentioned to me he had attended a talk by Aki Vehtari in Norway where my early work with Jérôme Dupuis on projective priors was used. He gave me the link to this paper by Peltola, Havulinna, Salomaa and Vehtari that indeed refers to the idea that a prior on a given Euclidean space defines priors by projections on all subspaces, despite the zero measure of all those subspaces. (This notion first appeared in a joint paper with my friend Costas Goutis, who alas died in a diving accident a few months later.) The projection further allowed for a simple expression of the Kullback-Leibler deviance between the corresponding models and for a Pythagorean theorem on the additivity of the deviances between embedded models. The weakest spot of this approach of ours was, in my opinion and unsurprisingly, about deciding when a submodel was too far from the full model. The lack of explanatory power introduced therein had no absolute scale and later discussions led me to think that the bound should depend on the sample size to ensure consistency. (The recent paper by Nott and Leng that was expanding on this projection has now appeared in CSDA.)

“Specifically, the models with subsets of covariates are found by maximizing the similarity of their predictions to this reference as proposed by Dupuis and Robert [12]. Notably, this approach does not require specifying priors for the submodels and one can instead focus on building a good reference model. Dupuis and Robert (2003) suggest choosing the size of the covariate subset based on an acceptable loss of explanatory power compared to the reference model. We examine using cross-validation based estimates of predictive performance as an alternative.” T. Peltola et al.

The paper also connects with the Bayesian Lasso literature, concluding on the horseshoe prior being more informative than the Laplace prior. It applies the selection approach to identify biomarkers with predictive performances in a study of diabetic patients. The authors rank model according to their (log) predictive density at the observed data, using cross-validation to avoid exploiting the data twice. On the MCMC front, the paper implements the NUTS version of HMC with STAN.