Pierre-Luc Bacon

Biography

Pierre-Luc Bacon is an assistant professor at Université de Montréal in the Department of Computer Science and Operations Research (DIRO). He is also a core academic member of Mila – Quebec Artificial Intelligence Institute and IVADO, and holds a Facebook CIFAR AI Chair. Bacon leads a research group that investigates the challenges posed by the curse of the horizon in reinforcement learning and optimal control.

Current Students

Mahdieh Adib

Collaborating researcher - Concordia University

Reza Alvandi

Research Intern - McGill University

Pablo Boitel

Collaborating researcher - ÉTS

Claire Bouillet

Research Intern - Polytechnique Montréal

Diego Calanzone

PhD - Université de Montréal

Kamen Damov

Professional Master's - Université de Montréal

Esther Derman

Collaborating Alumni - Université de Montréal

Co-supervisor :

Master's Research - Polytechnique Montréal

Principal supervisor :

Hanane Dagdougui

Arielle Gazzé

Master's Research - Université de Montréal

Adrien Goldszal

Collaborating Alumni - Université de Montréal

Niki Howe

PhD - Université de Montréal

Austin Jang

PhD - Université de Montréal

Akash Karthikeyan

PhD - Université de Montréal

Postdoctorate - McGill University

Principal supervisor :

Master's Research - Université de Montréal

Principal supervisor :

Yashar Hezaveh

Lu Li

PhD - Université de Montréal

Michel Ma

PhD - Université de Montréal

Vamsi Krishna Munjuluri V S

Artiom Matvei

Master's Research - Université de Montréal

Aneri Muni

PhD - Université de Montréal

Master's Research - Université de Montréal

Tianwei Ni

PhD - Université de Montréal

PhD - Université de Montréal

Gabriel Sasseville

PhD - Université de Montréal

Postdoctorate - Université de Montréal

Refrigerants are warming up the planet. AI can help.

Yihao Sun

PhD - Université de Montréal

Postdoctorate - Polytechnique Montréal

Postdoctorate - Université de Montréal

Principal supervisor :

Master's Research - Université de Montréal

Blog Posts

Diagram of the physics informed AI system

February 26, 2026

Adrien Goldszal

Pierre-Luc Bacon

Read the article

July 31, 2024

Neural Differential Equations for Temperature Control in Buildings Under Demand Response Programs

Vincent Taboga

Clement Gehring

Mathieu Le Cam

Hanane Dagdougui

Pierre-Luc Bacon

Read the article

August 31, 2022

Direct Behavior Specification via Constrained Reinforcement Learning

Julien Roy

Roger Girgis

Joshua Romoff

Pierre-Luc Bacon

Chris Pal

Read the article

Publications

State Entropy Regularization for Robust Reinforcement Learning

Yonatan Ashlag

Uri Koren

Mirco Mutti

Esther Derman

Shie Mannor

State entropy regularization has empirically shown better exploration and sample complexity in reinforcement learning (RL). However, its the… (see more)oretical guarantees have not been studied. In this paper, we show that state entropy regularization improves robustness to structured and spatially correlated perturbations. These types of variation are common in transfer learning but often overlooked by standard robust RL methods, which typically focus on small, uncorrelated changes. We provide a comprehensive characterization of these robustness properties, including formal guarantees under reward and transition uncertainty, as well as settings where the method performs poorly. Much of our analysis contrasts state entropy with the widely used policy entropy regularization, highlighting their different benefits. Finally, from a practical standpoint, we illustrate that compared with policy entropy, the robustness advantages of state entropy are more sensitive to the number of rollouts used for policy evaluation.

2025-09-17

NeurIPS.cc/2025/Conference (oral)

Scaling Trends in Language Model Robustness

Nikolaus Howe

Ian McKenzie

Oskar Hollinsworth

Michał Zając

Tom Tseng

Aaron Tucker

Adam Gleave

Increasing model size has unlocked a dazzling array of capabilities in modern language models. At the same time, even frontier models remain… (see more) vulnerable to jailbreaks and prompt injections, despite concerted efforts to make them robust. As both attack and defense gain access to more compute, and as models become larger, what happens to robustness? We argue that to answer this question requires a \emph{scaling} approach, which we employ in an extensive study of language model robustness across several classification tasks, model families, and adversarial attacks. We find that in the absence of explicit safety training, larger models are not consistently more robust; however, scale improves sample efficiency in adversarial training, though it worsens compute efficiency. Further, we find that increasing attack compute smoothly improves attack success rate against both undefended and adversarially trained models. Finally, after exploring robustness transfer across attacks and threat models, we combine attack and defense scaling rates to study the offense-defense balance. We find that while attack scaling outpaces adversarial training across all models studied, larger adversarially trained models might give defense the advantage in the long run. These results underscore the utility of the scaling lens, and provide a paradigm for evaluating future attacks and defenses on frontier models.

2025-07-14

International Conference on Machine Learning (Accept (spotlight poster))

proceedings.mlr.press

Discrete Compositional Generation via General Soft Operators and Robust Reinforcement Learning

Marco Jiralerspong

Esther Derman

Danilo Vucetic

Nikolay Malkin

Bilun Sun

Tianyu Zhang

Gauthier Gidel

A major bottleneck in scientific discovery consists of narrowing an exponentially large set of objects, such as proteins or molecules, to a … (see more)small set of promising candidates with desirable properties. While this process can rely on expert knowledge, recent methods leverage reinforcement learning (RL) guided by a proxy reward function to enable this filtering. By employing various forms of entropy regularization, these methods aim to learn samplers that generate diverse candidates that are highly rated by the proxy function. In this work, we make two main contributions. First, we show that these methods are liable to generate overly diverse, suboptimal candidates in large search spaces. To address this issue, we introduce a novel unified operator that combines several regularized RL operators into a general framework that better targets peakier sampling distributions. Secondly, we offer a novel, robust RL perspective of this filtering process. The regularization can be interpreted as robustness to a compositional form of uncertainty in the proxy function (i.e., the true evaluation of a candidate differs from the proxy's evaluation). Our analysis leads us to a novel, easy-to-use algorithm we name trajectory general mellowmax (TGM): we show it identifies higher quality, diverse candidates than baselines in both synthetic and real-world tasks. Code: https://github.com/marcojira/tgm.

2025-06-19

ArXiv (preprint)

arxiv.org

What Matters when Modeling Human Behavior using Imitation Learning?

As AI systems become increasingly embedded in human decision-making process, aligning their behavior with human values is critical to ensuri… (see more)ng safe and trustworthy deployment. A central approach to AI Alignment called Imitation Learning (IL), trains a learner to directly mimic desirable human behaviors from expert demonstrations. However, standard IL methods assume that (1) experts act to optimize expected returns; (2) expert policies are Markovian. Both assumptions are inconsistent with empirical findings from behavioral economics, according to which humans are (1) risk-sensitive; and (2) make decisions based on past experience. In this work, we examine the implications of risk sensitivity for IL and show that standard approaches do not capture all optimal policies under risk-sensitive decision criteria. By characterizing these expert policies, we identify key limitations of existing IL algorithms in replicating expert performance in risk-sensitive settings. Our findings underscore the need for new IL frameworks that account for both risk-aware preferences and temporal dependencies to faithfully align AI behavior with human experts.

2025-06-09

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

Understanding Behavioral Metric Learning: A Large-Scale Study on Distracting Reinforcement Learning Environments

Ziyan "Ray" Luo

A key approach to state abstraction is approximating behavioral metrics (notably, bisimulation metrics) in the observation space and embeddi… (see more)ng these learned distances in the representation space. While promising for robustness to task-irrelevant noise, as shown in prior work, accurately estimating these metrics remains challenging, requiring various design choices that create gaps between theory and practice. Prior evaluations focus mainly on final returns, leaving the quality of learned metrics and the source of performance gains unclear. To systematically assess how metric learning works in deep reinforcement learning (RL), we evaluate five recent approaches, unified conceptually as isometric embeddings with varying design choices. We benchmark them with baselines across 20 state-based and 14 pixel-based tasks, spanning 370 task configurations with diverse noise settings. Beyond final returns, we introduce the evaluation of a denoising factor to quantify the encoder's ability to filter distractions. To further isolate the effect of metric learning, we propose and evaluate an isolated metric estimation setting, in which the encoder is influenced solely by the metric loss. Finally, we release an open-source, modular codebase to improve reproducibility and support future research on metric learning in deep RL.

2025-05-08

rl-conference.cc/RLC/2025/Conference (accepted)

Network Sparsity Unlocks the Scaling Potential of Deep Reinforcement Learning

Guozheng Ma

Lu Li

Zilin Wang

Li Shen

Dacheng Tao

Effectively scaling up deep reinforcement learning models has proven notoriously difficult due to network pathologies during training, motiv… (see more)ating various targeted interventions such as periodic reset and architectural advances such as layer normalization. Instead of pursuing more complex modifications, we show that introducing static network sparsity alone can unlock further scaling potential beyond their dense counterparts with state-of-the-art architectures. This is achieved through simple one-shot random pruning, where a predetermined percentage of network weights are randomly removed once before training. Our analysis reveals that, in contrast to naively scaling up dense DRL networks, such sparse networks achieve both higher parameter efficiency for network expressivity and stronger resistance to optimization challenges like plasticity loss and gradient interference. We further extend our evaluation to visual and streaming RL scenarios, demonstrating the consistent benefits of network sparsity.

2025-04-30

ICML.cc/2025/Conference (oral)

proceedings.mlr.press

Mol-MoE: Training Preference-Guided Routers for Molecule Generation

Diego Calanzone

Pierluca D'Oro

Recent advances in language models have enabled framing molecule generation as sequence modeling. However, existing approaches often rely on… (see more) single-objective reinforcement learning, limiting their applicability to real-world drug design, where multiple competing properties must be optimized. Traditional multi-objective reinforcement learning (MORL) methods require costly retraining for each new objective combination, making rapid exploration of trade-offs impractical. To overcome these limitations, we introduce Mol-MoE, a mixture-of-experts (MoE) architecture that enables efficient test-time steering of molecule generation without retraining. Central to our approach is a preference-based router training objective that incentivizes the router to combine experts in a way that aligns with user-specified trade-offs. This provides improved flexibility in exploring the chemical property space at test time, facilitating rapid trade-off exploration. Benchmarking against state-of-the-art methods, we show that Mol-MoE achieves superior sample quality and steerability.

2025-03-04

ICLR.cc/2025/Workshop/GEM (published)

Maxwell's Demon at Work: Efficient Pruning by Leveraging Saturation of Neurons

When training neural networks, dying neurons -- units becoming inactive or saturated -- are traditionally seen as harmful. This paper sheds … (see more)new light on this phenomenon. By exploring the impact of various hyperparameter configurations on dying neurons during training, we gather insights on how to improve upon sparse training approaches to pruning. We introduce Demon Pruning (DemP), a method that controls the proliferation of dead neurons through a combination of noise injection on active units and a one-cycle schedule regularization strategy, dynamically leading to network sparsity. Experiments on CIFAR-10 and ImageNet datasets demonstrate that DemP outperforms existing dense-to-sparse structured pruning methods, achieving better accuracy-sparsity tradeoffs and accelerating training by up to 3.56

2025-02-12

TMLR (accepted)

MaestroMotif: Skill Design From Artificial Intelligence Feedback

Martin Klissarov

Mikael Henaff

Roberta Raileanu

Marlos C. Machado

Describing skills in natural language has the potential to provide an accessible way to inject human knowledge about decision-making into an… (see more) AI system. We present MaestroMotif, a method for AI-assisted skill design, which yields high-performing and adaptable agents. MaestroMotif leverages the capabilities of Large Language Models (LLMs) to effectively create and reuse skills. It first uses an LLM's feedback to automatically design rewards corresponding to each skill, starting from their natural language description. Then, it employs an LLM's code generation abilities, together with reinforcement learning, for training the skills and combining them to implement complex behaviors specified in language. We evaluate MaestroMotif using a suite of complex tasks in the NetHack Learning Environment (NLE), demonstrating that it surpasses existing approaches in both performance and usability.

2025-01-21

ICLR.cc/2025/Conference (oral)

Neural differential equations for temperature control in buildings under demand response programs

Mathieu Le Cam

2024-07-31

Applied Energy (published)

Do Transformer World Models Give Better Policy Gradients?

A natural approach for reinforcement learning is to predict future rewards by unrolling a neural network world model, and to backpropagate t… (see more)hrough the resulting computational graph to learn a policy. However, this method often becomes impractical for long horizons since typical world models induce hard-to-optimize loss landscapes. Transformers are known to efficiently propagate gradients over long horizons: could they be the solution to this problem? Surprisingly, we show that commonly-used transformer world models produce circuitous gradient paths, which can be detrimental to long-range policy gradients. To tackle this challenge, we propose a class of world models called Actions World Models (AWMs), designed to provide more direct routes for gradient propagation. We integrate such AWMs into a policy gradient framework that underscores the relationship between network architectures and the policy gradient updates they inherently represent. We demonstrate that AWMs can generate optimization landscapes that are easier to navigate even when compared to those from the simulator itself. This property allows transformer AWMs to produce better policies than competitive baselines in realistic long-horizon tasks.

2024-07-07

Proceedings of the 41st International Conference on Machine Learning (published)

proceedings.mlr.press

Bridging State and History Representations: Understanding Self-Predictive RL

Benjamin Eysenbach

Representations are at the core of all deep reinforcement learning (RL) methods for both Markov decision processes (MDPs) and partially obse… (see more)rvable Markov decision processes (POMDPs). Many representation learning methods and theoretical frameworks have been developed to understand what constitutes an effective representation. However, the relationships between these methods and the shared properties among them remain unclear. In this paper, we show that many of these seemingly distinct methods and frameworks for state and history abstractions are, in fact, based on a common idea of self-predictive abstraction. Furthermore, we provide theoretical insights into the widely adopted objectives and optimization, such as the stop-gradient technique, in learning self-predictive representations. These findings together yield a minimalist algorithm to learn self-predictive representations for states and histories. We validate our theories by applying our algorithm to standard MDPs, MDPs with distractors, and POMDPs with sparse rewards. These findings culminate in a set of preliminary guidelines for RL practitioners.

2024-01-15

ICLR.cc/2024/Conference (poster)