International conference
25-29th of October 2021, CIRM, Marseille
The design and analysis of robust Bayesian methods for high dimensional statistical settings is at the cornerstone of modern real world learning problems. They have recently attracted a lot of attention thanks to newly widespread massive computational resources and the keen interest for pivotal applications which require interpretable solutions and uncertainty estimation without which inference procedures provide very few guarantees.
The scientific core of this week is to provide a rich overview of end-to-end Bayesian solutions to real world learning problems in complex settings. The week will highlight all aspects of such approaches from data processing and feature engineering to Monte Carlo estimation procedures as well as the most recent theoretical guarantees obtained in high dimensional settings or frameworks with dependent and/or missing data.
The scientific program of this week includes 2 mini courses, 6 plenary talks, 15 contributed talks and a practical session.
The fully detailed program, containing also the abstracts of the poster session, is available here in the pdf
format.
Speaker | Title | ||
---|---|---|---|
8:45-9:00 | Introduction | ||
9:00-10:30 | Master Class | Rémi Bardenet | A tutorial on Bayesian machine learning: what, why and how ? |
11:00-11:30 | Contributed | Marina Riabiz | Optimal thinning of MCMC output |
11:30-12:00 | Contributed | Hai-Dang Dau | Waste-free Sequential Monte Carlo |
12:00-12:30 | Contributed | James Thornton | Differentiable particle filtering via entropy-regularized optimal transport |
14:00-15:00 | Invited talk | Anthony Lee | Intermediate distributions and complexity bounds for SMC |
15:30-16:30 | Invited talk | Judith Rousseau | Using cut posterior in semi parametric inference with applications to semiparametric and nonparametric Bayesian inference in hidden Markov models |
17:00-19:30 | Poster session |
Speaker | Title | ||
---|---|---|---|
9:00-10:30 | Master Class | Rémi Bardenet | A tutorial on Bayesian machine learning: what, why and how ? |
11:00-11:30 | Contributed | Takuo Matsubara | Robust generalised bayesian inference for intractable likelihoods |
11:30-12:00 | Contributed | Deborah Sulem | Bayesian estimation of nonlinear Hawkes process |
12:00-12:30 | Contributed | Torben Sell | Dimension-robust function space priors for stochastic control |
14:00-15:30 | Master Class | Elisabeth Gassiat | Bayesian multiple testing for dependent data and hidden Markov models |
16:00-17:00 | Invited talk | Florence Forbes | Simulation-based Bayesian inference for high dimensional inverse problems |
Speaker | Title | ||
---|---|---|---|
9:30-10:30 | Invited talk | Eric Moulines | TBA |
11:00-11:30 | Contributed | Bart Eggen | Bayesian senstivity analysis for a missing data model with continuous outcomes |
11:30-12:00 | Contributed | Gonzalo Mena | On the choice of priors for the estimation of infection fatality rates in the absence of serological data |
12:00-12:30 | Contributed | Solon Karapanagiotis | Tailored Bayes: a risk modelling framework under unequal misclassification costs |
14:00-19:30 | Social event |
Speaker | Title | ||
---|---|---|---|
9:30-11:00 | Master Class | Elisabeth Gassiat | Bayesian multiple testing for dependent data and hidden Markov models |
11:30-12:00 | Contributed | Amani Alahmadi | SMC ABC estimator for ODE models |
12:00-12:30 | Contributed | Luke Kelly | Coupled MCMC for Bayesian phylogenetic inference |
14:00-15:00 | Invited talk | Gareth Roberts | Regenerative non-reversible MCMC and the Restore algorithm |
15:30-16:30 | Invited talk | Randal Douc | The Kick Kac teleportation algorithm : boost your favorite MCMC using Kac formula |
18:00-19:30 | Social event |
Speaker | Title | ||
---|---|---|---|
9:30-10:00 | Contributed | Sebastiano Grazzi | Sticky PDMP samplers for sparce and local inference problems |
10:00-10:30 | Contributed | Andrea Bertazzi | Approximations of PDMP and their convergence properties |
10:30-11:00 | Contributed | Andi Wang | Subgeometric hypocoercivity for PDMP |
11:00-11:30 | Closing |
Vote for your favorite poster here!
To register for the ping pong tournament, click here.
It will be a double tournament, you’re partner will be randomly (with importance weights!) chosen.
The tournament will start on Tuesday!
Scientific committee
Organizing committee