Yoshua Bengio

Biographie

*Pour toute demande média, veuillez écrire à medias@mila.quebec.

Pour plus d’information, contactez Marie-Josée Beauchamp, adjointe administrative à marie-josee.beauchamp@mila.quebec.

Reconnu comme une sommité mondiale en intelligence artificielle, Yoshua Bengio s’est surtout distingué par son rôle de pionnier en apprentissage profond, ce qui lui a valu le prix A. M. Turing 2018, le « prix Nobel de l’informatique », avec Geoffrey Hinton et Yann LeCun. Il est professeur titulaire à l’Université de Montréal, fondateur et conseiller scientifique de Mila – Institut québécois d’intelligence artificielle, et codirige en tant que senior fellow le programme Apprentissage automatique, apprentissage biologique de l'Institut canadien de recherches avancées (CIFAR). Il occupe également la fonction de conseiller spécial et directeur scientifique fondateur d’IVADO.

En 2018, il a été l’informaticien qui a recueilli le plus grand nombre de nouvelles citations au monde. En 2019, il s’est vu décerner le prestigieux prix Killam. Depuis 2022, il détient le plus grand facteur d’impact (h-index) en informatique à l’échelle mondiale. Il est fellow de la Royal Society de Londres et de la Société royale du Canada, et officier de l’Ordre du Canada.

Soucieux des répercussions sociales de l’IA et de l’objectif que l’IA bénéficie à tous, il a contribué activement à la Déclaration de Montréal pour un développement responsable de l’intelligence artificielle.

Étudiants actuels

Jamal Abou Haibeh

Collaborateur·rice alumni - McGill

Mohammed Abukalam

Collaborateur·rice alumni - UdeM

Berkes Anaïs

Collaborateur·rice de recherche - Cambridge University

Superviseur⋅e principal⋅e :

Rim Assouel

Doctorat - UdeM

Stefan Bauer

Visiteur de recherche indépendant

Co-superviseur⋅e :

Guillaume Lajoie

Paul Bertin

Doctorat - UdeM

Joyce Chai

Visiteur de recherche indépendant

Superviseur⋅e principal⋅e :

Siva Reddy

Shahana Chatterjee

Collaborateur·rice de recherche - N/A

Superviseur⋅e principal⋅e :

Doctorat - UdeM

Collaborateur·rice de recherche - KAIST

Collaborateur·rice alumni - UdeM

Doctorat - UdeM

Collaborateur·rice alumni - UdeM

Co-superviseur⋅e :

Loubna Benabbou

Desmond Elliott

Visiteur de recherche indépendant

Superviseur⋅e principal⋅e :

Doctorat - UdeM

Co-superviseur⋅e :

Doctorat - UdeM

Doctorat - UdeM

Leo Feng

Doctorat - UdeM

Doctorat

Doctorat - UdeM

Edward Hu

Doctorat - UdeM

Moksh Jain

Doctorat - UdeM

Doctorat - UdeM

Superviseur⋅e principal⋅e :

Collaborateur·rice alumni - UdeM

Hyeonah Kim

Postdoctorat - UdeM

Superviseur⋅e principal⋅e :

Alex Hernandez-Garcia

Salem Lahlou

Collaborateur·rice alumni - UdeM

Tabitha Edith Lee

Postdoctorat - UdeM

Superviseur⋅e principal⋅e :

Zhen Liu

Collaborateur·rice alumni - UdeM

Superviseur⋅e principal⋅e :

Liam Paull

Chenghao Liu

Collaborateur·rice alumni

Doctorat - UdeM

Collaborateur·rice alumni - UdeM

Cristian Dragos Manta

Doctorat - UdeM

Co-superviseur⋅e :

Dhanya Sridhar

Sarthak Mittal

Doctorat - UdeM

Superviseur⋅e principal⋅e :

Doctorat - UdeM

Superviseur⋅e principal⋅e :

Postdoctorat - UdeM

Superviseur⋅e principal⋅e :

Visiteur de recherche indépendant - UdeM

Padideh Nouri

Doctorat - UdeM

Superviseur⋅e principal⋅e :

Ali Parviz

Collaborateur·rice de recherche - Ying Wu Coll of Computing

Lena Podina

Collaborateur·rice de recherche - University of Waterloo

Superviseur⋅e principal⋅e :

David Rolnick

Nassim Rahaman

Collaborateur·rice alumni - Max-Planck-Institute for Intelligent Systems

Amine RAZIG

Collaborateur·rice de recherche - UdeM

Co-superviseur⋅e :

Doctorat - UdeM

Postdoctorat - UdeM

Visiteur de recherche indépendant - UdeM

Oli RICHARDSON

Postdoctorat - UdeM

Camille Rochefort-Boulanger

Doctorat - UdeM

Superviseur⋅e principal⋅e :

Julie Hussin

Abhik Roychoudhury Roychoudhury

Visiteur de recherche indépendant

Superviseur⋅e principal⋅e :

Postdoctorat - UdeM

Collaborateur·rice alumni - UdeM

Marcin Sendera

Collaborateur·rice alumni - UdeM

Divya Sharma

Postdoctorat

Co-superviseur⋅e :

Alex Hernandez-Garcia

Marco Stock

Visiteur de recherche indépendant - Technical University of Munich

Mélisande Astrid Crystal Teng

Doctorat - UdeM

Co-superviseur⋅e :

Hugo Larochelle

Ivan Titov

Visiteur de recherche indépendant

Superviseur⋅e principal⋅e :

Siva Reddy

Alex Tong

Collaborateur·rice alumni - UdeM

Postdoctorat - UdeM

Co-superviseur⋅e :

Doctorat - UdeM

Superviseur⋅e principal⋅e :

Collaborateur·rice de recherche

Collaborateur·rice de recherche - UdeM

Skipper : combiner l’abstraction spatiale et temporelle afin d’améliorer la généralisation

Nicole Zhang

Doctorat - McGill

Superviseur⋅e principal⋅e :

Doctorat - UdeM

Superviseur⋅e principal⋅e :

Doctorat - UdeM

Harry Zhao

Collaborateur·rice alumni - McGill

Superviseur⋅e principal⋅e :

Billets de blogue

Generic thumbnail for Mila Blog articles.

22 février 2024

par

Mingde Harry Zhao

Safa Alver

Harm van Seijen

Romain Laroche

Doina Precup

Yoshua Bengio

Mise à l’échelle au service du raisonnement et de l’apprentissage automatique basé sur un modèle

Scaling in the service of reasoning & model-based ML

4 avril 2023

par

Yoshua Bengio

Edward J. Hu

Une collaboration entre Mila et Relation Therapeutics pour découvrir in vitro de nouvelles associations médicamenteuses synergiques

A collaboration between Mila and Relation Therapeutics to discover novel synergistic combinations of drugs in vitro

23 mars 2022

par

Paul Bertin

Jake P. Taylor-King

Yoshua Bengio

Les réseaux de flot génératifs

15 mars 2022

par

Yoshua Bengio

Publications

Low Compute Unlearning via Sparse Representations

Vedant Shah

Frederik Träuble

Ashish Malik

Hugo Larochelle

Michael Curtis Mozer

Sanjeev Arora

Anirudh Goyal

Machine unlearning, which involves erasing knowledge about a \emph{forget set} from a trained model, can prove to be costly and infeasible … (voir plus)using existing techniques. We propose a low-compute unlearning technique based on a discrete representational bottleneck. We show that the proposed technique efficiently unlearns the forget set and incurs negligible damage to the model's performance on the rest of the dataset. We evaluate the proposed technique on the problem of class unlearning using four datasets: CIFAR-10, CIFAR-100, LACUNA-100 and ImageNet-1k. We compare the proposed technique to SCRUB, a state-of-the-art approach which uses knowledge distillation for unlearning. Across all four datasets, the proposed technique performs as well as, if not better than SCRUB while incurring almost no computational cost.

2025-09-15

TMLR (accepté)

Relative Trajectory Balance is equivalent to Trust-PCL

Recent progress in generative modeling has highlighted the importance of Reinforcement Learning (RL) for fine-tuning, with KL-regularized me… (voir plus)thods in particular proving to be highly effective for both autoregressive and diffusion models. Complementing this line of work, the Relative Trajectory Balance (RTB) objective was recently introduced in the context of Generative Flow Networks (GFlowNets) to serve the same role of improving fine-tuning in sequential generative models. Building on prior work linking GFlowNets and maximum-entropy RL, we establish in this paper an equivalence between RTB and Trust-PCL, an off-policy RL method with KL regularization. This equivalence situates RTB within the broader theoretical landscape of KL-regularized RL, and clarifies its relationship to earlier methods. Leveraging this insight, we revisit an illustrative example from the RTB paper and show that KL-regularized RL methods achieve comparable performance, offering an alternative perspective to what was previously reported.

2025-09-01

arXiv (publié)

Relative Trajectory Balance is equivalent to Trust-PCL

2025-09-01

ArXiv (prépublication)

Relative Trajectory Balance is equivalent to Trust-PCL

2025-09-01

ArXiv (prépublication)

Torsional-GFN: a conditional conformation generator for small molecules

Alexandra Volokhova

Lena Nehale Ezzine

Piotr Gai'nski

Alex Hernandez-Garcia

Generating stable molecular conformations is crucial in several drug discovery applications, such as estimating the binding affinity of a mo… (voir plus)lecule to a target. Recently, generative machine learning methods have emerged as a promising, more efficient method than molecular dynamics for sampling of conformations from the Boltzmann distribution. In this paper, we introduce Torsional-GFN, a conditional GFlowNet specifically designed to sample conformations of molecules proportionally to their Boltzmann distribution, using only a reward function as training signal. Conditioned on a molecular graph and its local structure (bond lengths and angles), Torsional-GFN samples rotations of its torsion angles. Our results demonstrate that Torsional-GFN is able to sample conformations approximately proportional to the Boltzmann distribution for multiple molecules with a single model, and allows for zero-shot generalization to unseen bond lengths and angles coming from the MD simulations for such molecules. Our work presents a promising avenue for scaling the proposed approach to larger molecular systems, achieving zero-shot generalization to unseen molecules, and including the generation of the local structure into the GFlowNet model.

2025-07-15

ArXiv (prépublication)

RL, but don’t do anything I wouldn’t do

Michael K. Cohen

Marcus Hutter

Stuart Russell

In reinforcement learning (RL), if the agent’s reward differs from the designers’ true utility, even only rarely, the state distribution… (voir plus) resulting from the agent’s policy can be very bad, in theory and in practice. When RL policies would devolve into undesired behavior, a common countermeasure is KL regularization to a trusted policy ("Don’t do anything I wouldn’t do"). All current cutting-edge language models are RL agents that are KL-regularized to a "base policy" that is purely predictive. Unfortunately, we demonstrate that when this base policy is a Bayesian predictive model of a trusted policy, the KL constraint is no longer reliable for controlling the behavior of an advanced RL agent. We demonstrate this theoretically using algorithmic information theory, and while systems today are too weak to exhibit this theorized failure precisely, we RL-finetune a language model and find evidence that our formal results are plausibly relevant in practice. We also propose a theoretical alternative that avoids this problem by replacing the "Don’t do anything I wouldn’t do" principle with "Don’t do anything I mightn’t do".

2025-07-11

Proceedings of the Forty-first Conference on Uncertainty in Artificial Intelligence (publié)

proceedings.mlr.press

RL, but don't do anything I wouldn't do

Michael K. Cohen

Marcus Hutter

Stuart Russell

In reinforcement learning (RL), if the agent's reward differs from the designers' true utility, even only rarely, the state distribution res… (voir plus)ulting from the agent's policy can be very bad, in theory and in practice. When RL policies would devolve into undesired behavior, a common countermeasure is KL regularization to a trusted policy ("Don't do anything I wouldn't do"). All current cutting-edge language models are RL agents that are KL-regularized to a "base policy" that is purely predictive. Unfortunately, we demonstrate that when this base policy is a Bayesian predictive model of a trusted policy, the KL constraint is no longer reliable for controlling the behavior of an advanced RL agent. We demonstrate this theoretically using algorithmic information theory, and while systems today are too weak to exhibit this theorized failure precisely, we RL-finetune a language model and find evidence that our formal results are plausibly relevant in practice. We also propose a theoretical alternative that avoids this problem by replacing the "Don't do anything I wouldn't do" principle with "Don't do anything I mightn't do".

2025-07-11

Proceedings of the Forty-first Conference on Uncertainty in Artificial Intelligence (publié)

proceedings.mlr.press

Can a Bayesian Oracle Prevent Harm from an Agent?

Michael K. Cohen

Is there a way to design powerful AI systems based on machine learning methods that would satisfy probabilistic safety guarantees? With the … (voir plus)long-term goal of obtaining a probabilistic guarantee that would apply in every context, we consider estimating a context-dependent bound on the probability of violating a given safety specification. Such a risk evaluation would need to be performed at run-time to provide a guardrail against dangerous actions of an AI. Noting that different plausible hypotheses about the world could produce very different outcomes, and because we do not know which one is right, we derive bounds on the safety violation probability predicted under the true but unknown hypothesis. Such bounds could be used to reject potentially dangerous actions. Our main results involve searching for cautious but plausible hypotheses, obtained by a maximization that involves Bayesian posteriors over hypotheses. We consider two forms of this result, in the i.i.d. case and in the non-i.i.d. case, and conclude with open problems towards turning such theoretical results into practical AI guardrails.

2025-07-11

Proceedings of the Forty-first Conference on Uncertainty in Artificial Intelligence (publié)

proceedings.mlr.press

Discrete Feynman-Kac Correctors

Alan Aspuru-Guzik

The performance of Large Language Models (LLMs) directly depends on the size of the context that the model was trained on. Despite significa… (voir plus)nt progress in increasing the context size of the current models, some applications remain bottlenecked by the number of processed tokens at inference time. A particular mathematical problem LLMs can be used for is inferring parameters in a statistical model, given data-points as input. Here we make a case demonstrating that discrete diffusion models offer a promising avenue for scaling such parameter prediction tasks, by combining the outputs of the same model evaluated on different parts of the training data. We propose Discrete Fenyman-Kac Correctors --- a framework that allows for controlling the generated distribution of discrete masked diffusion models at inference time. We derive Sequential Monte Carlo (SMC) algorithms that, given a trained discrete diffusion model, sample from its annealed distribution or the product of distributions with different conditions. Notably, our framework does not require any training, finetuning and external reward functions. Finally, we apply our framework to amortized linear regression using LLaDA and demonstrate that it drastically outperforms the standard inference procedure in terms of accuracy and adherence to prompt format.

2025-07-09

ICML.cc/2025/Workshop/AI4MATH (poster)

HVAC-GRACE: Transferable Building Control via Heterogeneous Graph Neural Network Policies

Anaïs Berkes

Donna Vakalis

David Rolnick

Buildings consume 40% of global energy, with HVAC systems responsible for up to half of that demand. As energy use grows, optimizing HVAC ef… (voir plus)ficiency is critical to meeting climate goals. While reinforcement learning (RL) offers a promising alternative to rule-based control, real-world adoption is limited by poor sample efficiency and generalisation. We introduce HVAC-GRACE, a graph-based RL framework that models buildings as heterogeneous graphs and integrates spatial message passing directly into temporal GRU gates. This enables each zone to learn control actions informed by both its own history and its structural context. Our architecture supports zero-shot transfer by learning topology-agnostic functions—but initial experiments reveal that this benefit depends on sufficient conditioned zone connectivity to maintain gradient flow. These findings highlight both the promise and the architectural requirements of scalable, transferable RL for building control

2025-07-01

ICML.cc/2025/Workshop/CO-BUILD (poster)

The Singapore Consensus on Global AI Safety Research Priorities

Tegan Maharaj

Luke Ong

Stuart Russell

Dawn Song

Max Tegmark

Lan Xue

Ya-Qin Zhang

Stephen Casper

Wan Sie Lee

Sören Mindermann

Vanessa Wilfred

Vidhisha Balachandran

Fazl Barez

Michael Belinsky

Imane Bello

Malo Bourgon

Mark Brakel

Sim'eon Campos

Duncan Cass-Beggs … (voir 67 de plus)

Jiahao Chen

Rumman Chowdhury

Kuan Chua Seah

Jeff Clune

Juntao Dai

Agnès Delaborde

Nouha Dziri

Francisco Eiras

Joshua Engels

Jinyu Fan

Adam Gleave

Noah D. Goodman

Fynn Heide

Johannes Heidecke

Dan Hendrycks

Cyrus Hodes

Bryan Low Kian Hsiang

Minlie Huang

Sami Jawhar

Jingyu Wang

Adam Tauman Kalai

Meindert Kamphuis

Mohan S. Kankanhalli

Subhash Kantamneni

Mathias Bonde Kirk

Thomas Kwa

Jeffrey Ladish

Kwok-Yan Lam

Wan Lee Sie

Taewhi Lee

Xiaojian Li

Jiajun Liu

Chaochao Lu

Yifan Mai

Richard Mallah

Julian Michael

Nick Moës

Simon Möller

Kihyuk Nam

Kwan Yee Ng

Mark Nitzberg

Besmira Nushi

Sean O hEigeartaigh

Alejandro Ortega

Pierre Peigné

James Petrie

Benjamin Prud'homme

Reihaneh Rabbany

Nayat Sanchez-Pi

Sarah Schwettmann

Buck Shlegeris

Saad Siddiqui

Aradhana Sinha

Martín Soto

Cheston Tan

Dong Ting

William-Chandra Tjhi

Robert Trager

Brian Tse

H. AnthonyTungK.

John Willes

Denise Wong

W. Xu

Rongwu Xu

Yi Zeng

HongJiang Zhang

Djordje Zikelic

Rapidly improving AI capabilities and autonomy hold significant promise of transformation, but are also driving vigorous debate on how to en… (voir plus)sure that AI is safe, i.e., trustworthy, reliable, and secure. Building a trusted ecosystem is therefore essential – it helps people embrace AI with confidence and gives maximal space for innovation while avoiding backlash. This requires policymakers, industry, researchers and the broader public to collectively work toward securing positive outcomes from AI’s development. AI safety research is a key dimension. Given that the state of science today for building trustworthy AI does not fully cover all risks, accelerated investment in research is required to keep pace with commercially driven growth in system capabilities. Goals: The 2025 Singapore Conference on AI (SCAI): International Scientific Exchange on AI Safety aims to support research in this important space by bringing together AI scientists across geographies to identify and synthesise research priorities in AI safety. The result, The Singapore Consensus on Global AI Safety Research Priorities, builds on the International AI Safety Report-A (IAISR) chaired by Yoshua Bengio and backed by 33 governments. By adopting a defence-in-depth model, this document organises AI safety research domains into three types: challenges with creating trustworthy AI systems (Development), challenges with evaluating their risks (Assessment), and challenges with monitoring and intervening after deployment (Control). Through the Singapore Consensus, we hope to globally facilitate meaningful conversations between AI scientists and AI policymakers for maximally beneficial outcomes. Our goal is to enable more impactful R&D efforts to rapidly develop safety and evaluation mechanisms and foster a trusted ecosystem where AI is harnessed for the public good.

2025-06-25

ArXiv (prépublication)

Fast Monte Carlo Tree Diffusion: 100x Speedup via Parallel Sparse Planning

Jaesik Yoon

Hyeonseo Cho

Sungjin Ahn

Diffusion models have recently emerged as a powerful approach for trajectory planning. However, their inherently non-sequential nature limit… (voir plus)s their effectiveness in long-horizon reasoning tasks at test time. The recently proposed Monte Carlo Tree Diffusion (MCTD) offers a promising solution by combining diffusion with tree-based search, achieving state-of-the-art performance on complex planning problems. Despite its strengths, our analysis shows that MCTD incurs substantial computational overhead due to the sequential nature of tree search and the cost of iterative denoising. To address this, we propose Fast-MCTD, a more efficient variant that preserves the strengths of MCTD while significantly improving its speed and scalability. Fast-MCTD integrates two techniques: Parallel MCTD, which enables parallel rollouts via delayed tree updates and redundancy-aware selection; and Sparse MCTD, which reduces rollout length through trajectory coarsening. Experiments show that Fast-MCTD achieves up to 100x speedup over standard MCTD while maintaining or improving planning performance. Remarkably, it even outperforms Diffuser in inference speed on some tasks, despite Diffuser requiring no search and yielding weaker solutions. These results position Fast-MCTD as a practical and scalable solution for diffusion-based inference-time reasoning.

2025-06-11

ArXiv (prépublication)