Machine Learning workshop IN2P3-IRFU

Date de début :

28/09/2022 10:50

Date de fin :

28/09/2022 11:10

Lieu :

amphithéatre Pierre Gilles De Gennes - APC

Ville :

Paris

Producteur :

L'Institut national de physique nucléaire et de physique des particules

Durée :

0:15:16

Type :

video/mp4

Format :

mp4

Résolution :

1280x720

Neural Posterior Estimation with Differentiable Simulators

Simulation-Based Inference (SBI) is a promising Bayesian inference framework that alleviates the need for analytic likelihoods to estimate posterior distributions. Recent advances using neural density estimators in SBI algorithms have demonstrated the ability to achieve high-fidelity posteriors, at the expense of a large number of simulations ; which makes their application potentially very time-consuming when using complex physical simulations. In this work we focus on boosting the sample-efficiency of posterior density estimation using the gradients of the simulator. We present a new method to perform Neural Posterior Estimation (NPE) with a differentiable simulator. We demonstrate how gradient information helps constrain the shape of the posterior and improves sample-efficiency.

Justine Zeghal (APC)

Dernières vidéos

Du même auteur