Omiros Papaspiliopoulos (Università Bocconi): Bilinear mixed models and variational inference
Adrien Corenflos (University of Warwick): A coupling-based approach to f-divergences diagnostics for Markov chain Monte Carlo [recently discussed on the ‘Og]
Our local Mostly (and monthly) Monte Carlo seminar is back for a new academic year, now organized by Antoine Luciano and Timothy Johnston. The first session will take place at the PariSanté Campus on Friday 17 October 2025 (3:00pm, room 07), with the organisers opening the dance, with two talks:
3pm Timothy Johnston (CEREMADE, Université Paris Dauphine–PSL): Differential Privacy of Markov Chains
Joint work with Andrea Bertazzi, Alain Durmus and Gareth Roberts
In this talk we shall discuss differential privacy, a framework for quantifying the extent to which a random output depends on the information used to produce it. After introducing several related definition of differential privacy, we shall discuss techniques used to show the differential privacy of both trajectories and single draws from Markov Chains. In doing so we shall touch on a perturbation technique which allows for Wasserstein type bounds to be converted into stronger distances like the KL and Renyi divergence.
4pm Antoine Luciano (CEREMADE, Université Paris Dauphine–PSL): Permutations accelerate Approximate Bayesian Computation
Joint work with Charly Andral, Christian P. Robert and Robin J. Ryder
Approximate Bayesian Computation (ABC) methods have become essential tools for performing inference when likelihood functions are intractable or computationally prohibitive. However, their scalability remains a major challenge in hierarchical or high-dimensional models. In this paper, we introduce permABC, a new ABC framework designed for settings with both global and local parameters, where observations are grouped into exchangeable compartments. Building upon the Sequential Monte Carlo ABC (ABC-SMC) framework, permABC exploits the exchangeability of compartments through permutation-based matching, significantly improving computational efficiency. We then develop two further, complementary sequential strategies: Over Sampling, which facilitates early-stage acceptance by temporarily increasing the number of simulated compartments, and Under Matching, which relaxes the acceptance condition by matching only subsets of the data. These techniques allow for robust and scalable inference even in high-dimensional regimes. Through synthetic and real-world experiments – including a hierarchical Susceptible-Infectious-Recover model of the early COVID-19 epidemic across 94 French departments – we demonstrate the practical gains in accuracy and efficiency achieved by our approach.
Last week, I got COVID for the very first (detected) time! Although it took a while to realise this was COVID and not the common cold. It started with an overnight runny nose that I attributed to swimming outside the previous and cool morn, a minor temperature shock compared with the previous week in Guadeloupe. And did not spot the unusually high heart rate that night… The next afternoon, I was feeling a bit feverish (with no fever!) and definitely tired, but again attributed the feeling to faster bike rides to and from Paris Dauphine and the after-effect of Caribean jetlag. I then had a very long night and definitely felt okay the morning after (with nightly heart rate back to normal). It is only when my wife starting showing the same symptoms that I suspected more than a common cold. With confirmation from a COVID test two days later. By then I had unfortunately contaminated a few other people, all of whom recovered reasonably fast (phew!). I have since taken back to wear a face mask in any close encounter and warned my students. I wonder if I got it in the plane back or at Dauphine on the opening day. Definitely planning the continuation of my yearly vaccine next November!!!
Yesterday a arXival by Antoine Luciano, Charly Andral (both PhD students, now or then, at Paris Dauphine), Robin Ryder (formerly at Paris Dauphine, now at Imperial College London) and myself got posted. It proposes to improve the scalability of ABC methods by exploiting the (full or partial) exchangeability in the data by implementing permutation-based matching between observed and simulated samples. This significantly improves computational efficiency, which is further enhanced by sequential strategies such as over-sampling, which facilitates early-stage acceptance by temporarily increasing the number of simulated compartments, and under-matching, which relaxes the acceptance condition by matching only subsets of the data. The map of France appears in connection with an application of the method to estimating SIR parameters, department by department. (It is also reminding me of the cover of Markov Chain Monte Carlo methods in practice, the 1996 contributed book edited by Wally Gilks, Sylvia Richardson and David Spiegelhalter.)
The main BayesComp²⁵ conference started with Pierre Jacob’s plenary talk on his recent advances on coupling for unbiased MCMC—currently ERC grantee on that topic—. Raising lazy questions like using a different target or transition kernel for the second chain in the coupling, connecting the Poisson equation and control variates, handling the signed issue with the unbiased approximations. Interestingly, they obtain an unbiased estimator of the asymptotic variance of the unbiased estimator. And a correction for self-normalised importance sampling, which has some connections with our 1996 (?) pinball sampler. Also an evaluation of the median of means, rather than the average of means, which is a thing I had been (lazily) contemplating for a while (On the greedy side, as I was writing my recovery exam for my Monte Carlo course, I realised the results Pierre presented could be somewhat recycled into exam problems!)
My first parallel session was on gradient-based methods with a talk by Francesca Crucinio on proximal particle Langevin algorithms (similar to the one she gave in PariSanté last year), a talk by Zhihao Wang on stereographic multiple try Metropolis(-Rosenbluth-Teller) that unsurprisingly recovers ergodicity thanks to the compactness of the ball. For which I wonder why a Normal proposal makes complete sense since one could consider a mover after the projection instead and why iid rather than repelling multiple proposals are used… The last speaker was just out from the plane from California, Siddharth Vishwanath who spoke about repelling-attracting HMC. With very nice animations of HMC, if reaching the main point of using both negative and positive frictions a few minutes before the session finished. The method preserves volume and potential, if not energy.
Speaking of which (energy), I find myself struggling with my less than 6 hours of sleep since arrival during the first afternoon session, despite a fiery hot spot lunch, which means in plainer terms that I alas dozed in and out of the talks. The second session saw Jack Jewson exposing in deeper details the exact PDMP algorithm for Gibbs measures Jeremias Knoblauch mentioned yesterday. And Jonathan Huggins as well, using Gaussian processes as proxies for expected likelihoods, with lower guarantees than pseudo-marginal versions. In a mildly connected way, Robin Ryder went through the resolution of the ecological inference challenge they produce with Nicolas Chopin and Théo Valdoire (all authors with whom I am connected, Théo being a brillant student of our MASH Master last year and now in Harvard, hopefully till the end of his PhD!)
On the extra-academic curriculum, I had a yummy dinner in the Maxwell Hawker (street) food centre, incl. Xiao Long Bao that cooled down fast enough to avoid the usual scalding effect, plus rojak a mixed fruit and vegetable fried in a peanut sauce that I had never tasted before, popiah (ditto), chili noodles, and an appam with durian deepfried balls as a fabulous and unexpected dessert.
 | 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… |