a chance (?) encounter

As I was cycling to Paris Dauphine, a few days ago, I spotted someone sitting on a bench and working on a laptop who suspiciously looked like… Andrew Gelman! As I knew Andrew was in Paris that week, and as we were reasonably close to Dauphine, this did not sound like a zero probability event. I thus stopped to check that indeed this was the real Andrew, who happened to be in the vicinity and had decided to run this double blind experiment as to whether or not we could spot one another. While I am reasonably aware of my surroundings when cycling (as a matter of mere survival), my radar rarely extends to people sitting on benches, especially when I am riding the middle white line on the boulevard. As I was further a wee bit late that day, I should have been in my office by the time Andrew sat there. A chance encounter, hence, or a super subjective inference from the author of BDA!

ABC with Gibbs steps

With Grégoire Clarté, Robin Ryder and Julien Stoehr, all from Paris-Dauphine, we have just arXived a paper on the specifics of ABC-Gibbs, which is a version of ABC where the generic ABC accept-reject step is replaced by a sequence of n conditional ABC accept-reject steps, each aiming at an ABC version of a conditional distribution extracted from the joint and intractable target. Hence an ABC version of the standard Gibbs sampler. What makes it so special is that each conditional can (and should) be conditioning on a different statistic in order to decrease the dimension of this statistic, ideally down to the dimension of the corresponding component of the parameter. This successfully bypasses the curse of dimensionality but immediately meets with two difficulties. The first one is that the resulting sequence of conditionals is not coherent, since it is not a Gibbs sampler on the ABC target. The conditionals are thus incompatible and therefore convergence of the associated Markov chain becomes an issue. We produce sufficient conditions for the Gibbs sampler to converge to a stationary distribution using incompatible conditionals. The second problem is then that, provided it exists, the limiting and also intractable distribution does not enjoy a Bayesian interpretation, hence may fail to be justified from an inferential viewpoint. We however succeed in producing a version of ABC-Gibbs in a hierarchical model where the limiting distribution can be explicited and even better can be weighted towards recovering the original target. (At least with limiting zero tolerance.)

logicomix redux

I had not made the link until the last speaker of the 50 years of Dauphine commemoration was introduced that he was one of the authors of Logicomix. He spoke of the mathematical modeling of neurons and brain activity, rather than comics, but at a very low level that he called cartoonesque. It is a rare event that cartoon characters can be met in the flesh!

postdoc position still open

The post-doctoral position supported by the ANR funding of our Paris-Saclay-Montpellier research conglomerate on approximate Bayesian inference and computation remains open for the time being. We are more particularly looking for candidates with a strong background in mathematical statistics, esp. Bayesian non-parametrics, towards the analysis of the limiting behaviour of approximate Bayesian inference. Candidates should email me (gmail address: bayesianstatistics) with a detailed vita (CV) and a motivation letter including a research plan. Letters of recommendation may also be emailed to the same address.

instituts interdisciplinaires d’intelligence artificielle

Four French institutes have just been selected by [the jury mandated by] the French Government to become interdisciplinary institutes for artificial intelligence (3IA):

•    ANITI (Toulouse), centred at l’Université Fédérale Toulouse Midi-Pyrénées
•    MIAI @ Grenoble Alpes (Grenoble), centred at l’Université Grenoble Alpes
•    PRAIRIE (Paris), centred à Inria
•    3IA Côte d’Azur (Nice), centred at l’Université Côte d’Azur

with Paris Dauphine, PSL Research University, a component of the third institute. (Where Prairie stands for PaRis Artificial Intelligence Research InstitutE.) This follows the “plan Villani“, launched by Cédric Villani to boost research in artificial intelligence in France and [possibly] slow down the brain drain of French IA researchers. With individual chairs attribued to each project, soon to be announced.

heart of Paris

Notre Drame