A bunch of ABC papers on arXiv yesterday, most of them linked to the incoming Handbook of ABC:
I remember fondly the early Valencia meetings where we did not have to pick between sessions. Then one year there were two sessions and soon more. And we now have to pick among equally tantalising sessions. [Complaint of the super wealthy, I do realise.] After a morning trip to San’Antioco and the southern coast of Sardinia, I started my ISBA 2016 with an not [that Bayesian] high dimension session with Michael Jordan (who gave a talk related to his MCMski lecture), Isa Verdinelli and Larry Wasserman.
Larry gave a [non-Bayesian, what else?!] talk on the problem of data splitting versus double use of the same data. Or rather using a model index estimated from a given dataset to estimate the properties of the mean of the same data. As in model selection. While splitting the data avoids all sorts of problem, not splitting the data but using a different loss function could avoid the issue. (And the infinite regress that if we keep conducting inference, we may have to split further and further the data.) Namely, if we were looking only at quantities that do not vary across models. So it is surprising that prediction get affected by this.
In a second session around Bayesian tests and model choice, Sarah Filippi presented the Bayesian non-parametric test she devised with Chris Holmes, using Polya trees. And mentioned our testing-by-mixture approach as a valuable alternative! Veronika Rockova talked about her new approach to efficient variable selection by spike-and-slab priors, through a mix of particle MCMC and EM, plus some variational Bayes motivations. (She also mentioned extensions by repulsive sampling through the pinball sampler, of which her recent AISTATS paper reminded me.)
Later in the evening, I figured out that the poster sessions that make the ISBA/Valencia meetings so unique are alas out of reach for me as the level of noise and my reduced hearing capacities (!) make impossible any prolonged discussion on any serious notion. No poster session for ‘Og’s men!, then, even though I can hang out at the fringe and chat with friends!
Simon Barthelmé gave his mini-course on EP, with loads of details on the implementation of the method. Focussing on the EP-ABC and MCMC-EP versions today. Leaving open the difficulty of assessing to which limit EP is converging. But mentioning the potential for asynchronous EP (on which I would like to hear more). Ironically using several times a logistic regression example, if not on the Pima Indians benchmark! He also talked about approximate EP solutions that relate to consensus MCMC. With a connection to Mark Beaumont’s talk at NIPS [at the time as mine!] on the comparison with ABC. While we saw several talks on EP during this week, I am still agnostic about the potential of the approach. It certainly produces a fast proxy to the true posterior and hence can be exploited ad nauseam in inference methods based on pseudo-models like indirect inference. In conjunction with other quick and dirty approximations when available. As in ABC, it would be most useful to know how far from the (ideal) posterior distribution does the approximation stands. Machine learning approaches presumably allow for an evaluation of the predictive performances, but less so for the modelling accuracy, even with new sampling steps. [But I know nothing, I know!]
Dennis Prangle presented some on-going research on high dimension [data] ABC. Raising the question of what is the true meaning of dimension in ABC algorithms. Or of sample size. Because the inference relies on the event d(s(y),s(y’))≤ξ or on the likelihood l(θ|x). Both one-dimensional. Mentioning Iain Murray’s talk at NIPS [that I also missed]. Re-expressing as well the perspective that ABC can be seen as a missing or estimated normalising constant problem as in Bornn et al. (2015) I discussed earlier. The central idea is to use SMC to simulate a particle cloud evolving as the target tolerance ξ decreases. Which supposes a latent variable structure lurking in the background.
Judith Rousseau gave her talk on non-parametric mixtures and the possibility to learn parametrically about the component weights. Starting with a rather “magic” result by Allman et al. (2009) that three repeated observations per individual, all terms in a mixture are identifiable. Maybe related to that simpler fact that mixtures of Bernoullis are not identifiable while mixtures of Binomial are identifiable, even when n=2. As “shown” in this plot made for X validated. Actually truly related because Allman et al. (2009) prove identifiability through a finite dimensional model. (I am surprised I missed this most interesting paper!) With the side condition that a mixture of p components made of r Bernoulli products is identifiable when p ≥ 2[log² r] +1, when log² is base 2-logarithm. And [x] the upper rounding. I also find most relevant this distinction between the weights and the remainder of the mixture as weights behave quite differently, hardly parameters in a sense.
This morning, we had a jam session at the maths department of Paris-Dauphine where a few researchers & colleagues of mine presented their field of research to the whole department. Very interesting despite or thanks to the variety of topics, with forays into the three-body problem(s) [and Poincaré‘s mistake], mean fields for Nash equilibrium (or how to exit a movie theatre), approximate losses in machine learning and so on. Somehow, there was some unity as well through randomness, convexity and optimal transport. One talk close to my own interests was obviously the study of simulation within convex sets by Joseph Lehec from Paris-Dauphine [and Sébastien Bubeck & Ronen Eldan] as they had established a total variation convergence result at a speed only increasing polynomially with the dimension. The underlying simulation algorithm is rather theoretical in that it involves random walk (or Langevin corrected) moves where any excursion outside the convex support is replaced with its projection on the set. Projection that may prove pretty expensive to compute if the convex set is defined for instance as the intersection of many hyperplanes. So I do not readily see how the scheme can be recycled into a competitor to a Metropolis-Hastings solution in that the resulting chain hits the boundary from time to time. With the same frequency over iterations. A solution is to instead use Metropolis-Hastings of course, while another one is to bounce on the boundary and then correct by Metropolis-Hastings… The optimal scales in the three different cases are quite different, from √d in the Metropolis-Hastings cases to d√d in the projection case. (I did not follow the bouncing option to the end, as it lacks a normalising constant.) Here is a quick and not particularly helpful comparison of the exploration patterns of both approaches in dimension 50 for the unit sphere and respective scales of 10/d√d [blue] and 1/√d [gold].
When discussing high dimension models with Ingmar Schüster Schuster [blame my fascination for accented characters!] the other day, we came across the following paradox with Metropolis-Hastings algorithms. If attempting to simulate from a multivariate standard normal distribution in a large dimension, when starting from the mode of the target, i.e., its mean γ, leaving the mode γis extremely unlikely, given the huge drop between the value of the density at the mode γ and at likely realisations (corresponding to the blue sequence). Even when relying on the very scale that makes the proposal identical to the target! Resorting to a tiny scale like Σ/p manages to escape the unhealthy neighbourhood of the highly unlikely mode (as shown with the brown sequence).
Here is the corresponding R code:
[sourcecode language=”r” gutter=”false”]
p=100
T=1e3
mh=mu #mode as starting value
vale=rep(0,T)
for (t in 1:T){
prop=mvrnorm(1,mh,sigma/p)
if (log(runif(1))<logdmvnorm(prop,mu,sigma)-
logdmvnorm(mh,mu,sigma)) mh=prop
vale[t]=logdmvnorm(mh,mu,sigma)}
[/sourcecode]
 | David on Bayesian workflow [book review… |
 | Ewan on Tidy up, Claude! |
 | xi'an on Tidy up, Claude! |
| Ewan on Tidy up, Claude! |
| Ewan on Já við viðræðum. Já við Evrópu… |