French Spring School in Theoretical Computer Science / École de Printemps d'Informatique Théorique

Collection French Spring School in Theoretical Computer Science / École de Printemps d'Informatique Théorique

Organisateur(s) Baudart, Guillaume ; Pagani, Michele ; Petrisan, Daniela ; Tasson, Christine

Date(s) 11/05/2026 - 15/05/2026

URL associée https://conferences.cirm-math.fr/3489.html

00:00:00 / 00:00:00

5 5

Ensuring correctness of probabilistic inference by static analysis

De Xavier Rival

Probabilistic inference engines lie at the core of probabilistic programming languages. However, their correctness depends on complex mathematical properties, that many probabilistic models do not satisfy. In this lecture, we will consider a family of inference engines based on variational inference and study properties required to ensure their correctness. After describing stochastic variational inference (SVI) and showing a few common pitfalls, we will set up a static analysis approach based on a rigorous study of the semantics of the inference engine. We will identify several key semantic properties of probabilistic models and show how they can be computed statically using abstract interpretation-based static analysis. Semantic properties under consideration will include consistency of distribution sampling and various smoothness properties.To address the static analysis part, we will introduce all required background in semantics, abstract interpretation and static analysis.

Informations sur la vidéo

Date de captation 12/05/2026
Date de publication 26/05/2026
Institut CIRM
Licence CC BY NC ND
Langue Anglais
Audience Chercheurs, Doctorants
Réalisateur(s) Luca Recanzone
Format MP4

Données de citation

DOI 10.24350/CIRM.V.20480903
Citer cette vidéo Rival, Xavier (12/05/2026). Ensuring correctness of probabilistic inference by static analysis. CIRM. Audiovisual resource. DOI: 10.24350/CIRM.V.20480903
URL https://dx.doi.org/10.24350/CIRM.V.20480903

Domaine(s)

Programming Languages

Bibliographie

LEE, Wonyeol, YU, Hangyeol, RIVAL, Xavier, et al. Towards verified stochastic variational inference for probabilistic programs. Proceedings of the ACM on Programming Languages, 2019, vol. 4, no POPL, p. 1-33. - https://doi.org/10.1145/3371084

Codes MSC

Document(s)

https://www.cirm-math.fr/RepOrga/3489/Slides/epit-ver-svi-0-intro.pdf

Dernières questions liées sur MathOverflow

Pour poser une question, votre compte Carmin.tv doit être connecté à mathoverflow

Poser une question sur MathOverflow

Toutes les vidéos de la collection

01:27:45

publiée le 26 mai 2026

Applicative bisimulation for higher-order probabilistic languages

De Raphaëlle Crubillé

01:14:40

publiée le 26 mai 2026

Expressive probabilistic programming : discrete-time stochastic processes

De Sam Staton

01:23:06

publiée le 26 mai 2026

Introduction to probabilistic programming

De Christine Tasson

01:24:49

publiée le 26 mai 2026

Algorithmic foundations for exact discrete probabilistic reasoning

De Steven Holtzen

01:26:21

publiée le 26 mai 2026

Ensuring correctness of probabilistic inference by static analysis

De Xavier Rival

Copyright Carmin.tv 2026

Donner son avis