EM degeneracy

At the MHC 2021 conference today (to which I biked to attend for real!, first time since BayesComp!) I listened to Christophe Biernacki exposing the dangers of EM applied to mixtures in the presence of missing data, namely that the algorithm has a rising probability to reach a degenerate solution, namely a single observation component. Rising in the proportion of missing data. This is not hugely surprising as there is a real (global) mode at this solution. If one observation components are prohibited, they should not be accepted in the EM update. Just as in Bayesian analyses with improper priors, the likelihood should bar single or double  observations components… Which of course makes EM harder to implement. Or not?! MCEM, SEM and Gibbs are obviously straightforward to modify in this case.

Judith Rousseau also gave a fascinating talk on the properties of non-parametric mixtures, from a surprisingly light set of conditions for identifiability to posterior consistency . With an interesting use of several priors simultaneously that is a particular case of the cut models. Namely a correct joint distribution that cannot be a posterior, although this does not impact simulation issues. And a nice trick turning a hidden Markov chain into a fully finite hidden Markov chain as it is sufficient to recover a Bernstein von Mises asymptotic. If inefficient. Sylvain LeCorff presented a pseudo-marginal sequential sampler for smoothing, when the transition densities are replaced by unbiased estimators. With connection with approximate Bayesian computation smoothing. This proves harder than I first imagined because of the backward-sampling operations…

transport Monte Carlo

Read this recent arXival by Leo Duan (from UF in Gainesville) on transport approaches to approximate Bayesian computation, in connection with normalising flows. The author points out a “lack of flexibility in a large class of normalizing flows”  to bring forward his own proposal.

“…we assume the reference (a multivariate uniform distribution) can be written as a mixture of many one-to-one transforms from the posterior”

The transportation problem is turned into defining a joint distribution on (β,θ) such that θ is marginally distributed from the posterior and β is one of an infinite collection of transforms of θ. Which sounds quite different from normalizing flows, to be sure. Reverting the order, if one manages to simulate β from its marginal the resulting θ is one of the transforms. Chosen to be a location-scale modification of β, s⊗β+m. The weights when going from θ to β are logistic transforms with Dirichlet distributed scales. All with parameters to be optimised by minimising the Kullback-Leibler distance between the reference measure on β and its inverse mixture approximation, and resorting to gradient descent. (This may sound a wee bit overwhelming as an approximation strategy and I actually had to make a large cup of strong macha to get over it, but this may be due to the heat wave occurring at the same time!) Drawing θ from this approximation is custom-made straightforward and an MCMC correction can even be added, resulting in an independent Metropolis-Hastings version since the acceptance ratio remains computable. Although this may defeat the whole purpose of the exercise by stalling the chain if the approximation is poor (hence suggesting this last step being used instead as a control.)

The paper also contains a theoretical section that studies the approximation error, going to zero as the number of terms in the mixture, K, goes to infinity. Including a Monte Carlo error in log(n)/n (and incidentally quoting a result from my former HoD at Paris 6, Paul Deheuvels). Numerical experiments show domination or equivalence with some other solutions, e.g. being much faster than HMC, the remaining $1000 question being of course the on-line evaluation of the quality of the approximation.

stratified ABC [One World ABC webinar]

The third episode of the One World ABC seminar (Season 1!) was kindly delivered by Umberto Picchini on Stratified sampling and bootstrapping for ABC which I already if briefly discussed after BayesComp 2020. Which sounds like a million years ago… His introduction on the importance of estimating the likelihood using a kernel, while 600% justified wrt his talk, made the One World ABC seminar sounds almost like groundhog day!  The central argument is in the computational gain brought by simulating a single θ dependent [expensive] dataset followed by [cheaper] bootstrap replicates. Which turns de fact into bootstrapping the summary statistics.

If I understand correctly, the post-stratification approach of Art Owen (2013?, I cannot find the reference) corrects a misrepresentation of mine. Indeed, defining a partition with unknown probability weights seemed to me to annihilate the appeal of stratification, because the Bernoulli variance of the estimated probabilities brought back the same variability as the mother estimator. But with bootstrap, this requires only two simulations, one for the weights and one for the target. And further allows for a larger ABC tolerance in fine. Free lunch?!

The speaker in two weeks (21 May or Ascension Thursday!) is my friend and co-author Gael Martin from Monash University, who will speak on Focused Bayesian prediction, at quite a late time down under..!

off to BayesComp 20, Gainesville

estimating the marginal likelihood (or an information criterion)

Tory Imai (from Kyoto University) arXived a paper last summer on what first looked like a novel approximation of the marginal likelihood. Based on the variance of thermodynamic integration. The starting argument is that there exists a power 0<t⁰<1 such that the expectation of the logarithm of the product of the prior by the likelihood to the power t⁰ or t⁰-powered likelihood  is equal to the standard log-marginal

when the expectation is under the posterior corresponding to the t⁰-powered likelihood (rather than the full likelihood). By an application of the mean value theorem. Watanabe’s (2013) WBIC replaces the optimum t⁰ with 1/log(n), n being the sample size. The issue in terms of computational statistics is of course that the error of WBIC (against the true log m(x)) is only characterised as an order of n.

The second part of the paper is rather obscure to me, as the motivation for the real log canonical threshold is missing, even though the quantity is connected with the power likelihood. And the DIC effective dimension. It then goes on to propose a new approximation of sBIC, where s stands for singular, of Drton and Plummer (2017) which I had missed (and may ask my colleague Martin later today at Warwick!). Quickly reading through the later however brings explanations about the real log canonical threshold being simply the effective dimension in Schwarwz’s BIC approximation to the log marginal,

(as derived by Watanabe), where m is called the multiplicity of the real log canonical threshold. Both λ and m being unknown, Drton and Plummer (2017) estimate the above approximation in a Bayesian fashion, which leads to a double indexed marginal approximation for a collection of models. Since this thread leads me further and further from a numerical resolution of the marginal estimation, but brings in a different perspective on mixture Bayesian estimation, I will return to this highly  in a later post. The paper of Imai discusses a different numerical approximation to sBIC, With a potential improvement in computing sBIC. (The paper was proposed as a poster to BayesComp 2020, so I am looking forward discussing it with the author.)