Doina Precup

busra-tugce.gurbuz@mila.quebec

Jesse Farebrother

Doctorat - McGill

Superviseur⋅e principal⋅e :

Marc Gendron-Bellemare

Doctorat - McGill

Superviseur⋅e principal⋅e :

Eilif B. Muller

Collaborateur·rice de recherche - Birla Institute of Technology

herambpatil2004@gmail.com

Mohammad Sami Nur Islam Islam

Jonathan Hu

Maîtrise recherche - McGill

jonathan.hu@mila.quebec

Howard Huang

Doctorat - McGill

huanghow@mila.quebec

Haque Ishfaq

Collaborateur·rice alumni - McGill

Maîtrise recherche - McGill

mohammad-sami-nur.islam@mila.quebec

flemming.kondrup@mila.quebec

Hangzhan Jin

Doctorat - Polytechnique

jinhangzhan@gmail.com

Doctorat - McGill

Postdoctorat - McGill

Jonathan Lebensold

Collaborateur·rice alumni - McGill

Collaborateur·rice alumni - McGill

jeremy.lucas@mila.quebec

Ray Luo

Doctorat - McGill

Superviseur⋅e principal⋅e :

Xujie Si

luo.ziyan@mila.quebec

Site web

Sicheng Lyu

Maîtrise recherche - McGill

sicheng.lyu@mila.quebec

G McCracken

Doctorat - McGill

mccrackg@mila.quebec

Nazanin Mohammadi Sepahvand

Collaborateur·rice alumni - McGill

sepahvan@mila.quebec

Shahrad Mohammadzadeh

Maîtrise recherche - McGill

Superviseur⋅e principal⋅e :

Reihaneh Rabbany

shahrad.mohammadzadeh@mila.quebec

Site web

padideh.nouri@mila.quebec

Gabriela Moisescu-Pareja

Collaborateur·rice de recherche - McGill

Co-superviseur⋅e :

Irina Rish

moisescg@mila.quebec

Padideh Nouri

Doctorat - UdeM

Co-superviseur⋅e :

Yoshua Bengio

Doctorat - McGill

Co-superviseur⋅e :

Stagiaire de recherche - McGill

william.potter@mila.quebec

manoosh.samiei@mila.quebec

Nate Rahn

Doctorat - McGill

Superviseur⋅e principal⋅e :

Marc Gendron-Bellemare

nathan.rahn@mila.quebec

Manoosh Samiei

Doctorat - McGill

Co-superviseur⋅e :

Paul Masset

Doctorat - McGill

Co-superviseur⋅e :

Blake Richards

samiemandana@gmail.com

Doctorat - McGill

Nishanth Anand Vemgal

Doctorat - McGill

Doctorat - McGill

Co-superviseur⋅e :

Samira Ebrahimi Kahou

Stagiaire de recherche - McGill

guangyuan.wang@mila.quebec

Zihan Wang

Doctorat - McGill

zihan.wang@mila.quebec

Site web

shuyuan.zhang@mila.quebec

Steve Wen

Maîtrise recherche - McGill

Co-superviseur⋅e :

Gregory Dudek

steve.wen@mila.quebec

Zijing Wu

Doctorat - McGill

Superviseur⋅e principal⋅e :

Paul Masset

zijing.wu@mila.quebec

Doctorat - McGill

Harry Zhao

Collaborateur·rice alumni - McGill

Co-superviseur⋅e :

Billets de blogue

Generic thumbnail for Mila Blog articles.

22 février 2024

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

par

Mingde Harry Zhao

Safa Alver

Harm van Seijen

Romain Laroche

Doina Precup

Yoshua Bengio

Lire l'article

Publications

Adaptive Multi-Horizon Reinforcement Learning

Manoosh Samiei

Paul Masset

Effective decision-making in complex and changing environments requires balancing short-term and long-term consequences. In reinforcement le… (voir plus)arning (RL), this trade-off is typically controlled through a fixed discount factor, which imposes a single exponentially discounted temporal horizon. However, biological agents exhibit flexible and adaptive temporal discounting, suggesting that effective planning requires multiple timescales. Here, we propose a multi-horizon approach that adaptively selects and combines temporal horizons, enabling robust adaptation to changes in reward structure without manual discount-factor tuning. This flexibility makes the method particularly suitable for continual learning scenarios involving task switches and varying environmental configurations. Empirically, we demonstrate that our approach identifies effective discount factors across a range of MiniGrid environments, including continual settings composed of three sequentially changing tasks. These results suggest that adaptive temporal discounting can improve parameter efficiency and enhance adaptability in both artificial and biologically inspired learning systems.

2026-07-21

arXiv (prépublication)

Addressing Sparse-Rewards in RL with Scalable Hierarchical Novel Eigen Options

Isabeau Prémont-Schwarz

Temporally extended exploration via graph Laplacian-based options is a promising approach to sparse-reward reinforcement learning (RL), but … (voir plus)existing methods either do not explicitly target novelty or fail to scale to pixel-based domains under function approximation. Novel Exploration via Orthogonality (NEO) addresses the first issue by constructing options that navigate from highly visited regions toward less visited ones, yet prior results were limited to settings where exact eigenvectors can be computed. We present a scalable extension of NEO to pixel-based domains, built on three contributions. First, we use a novelty-weighted continuous Laplacian graph-drawing objective, which enables RL with continuous observations. Second, we embed the resulting eigen-potential options within a hierarchical reinforcement learning framework, enabling coherent temporally extended behavior. Third, we observe that learned eigen-potential rewards are directional but locally unreliable under online approximation; we therefore augment each option reward with a novelty bonus, a novel design idea that proves essential for stabilizing option learning while preserving novelty-directed exploration. Together, these contributions yield stronger and more persistent exploration, enabling longer option rollouts and better access to hard-to-reach novel states. Empirically, our method significantly outperforms both the prior scalable Laplacian-option baseline and a direct extension of NEO on sparse-reward benchmarks under a fixed budget. On Montezuma's Revenge, our best variant achieves approximately 1.8x higher return than both baselines. On Venture, both baselines yield returns near zero, whereas our method achieves a return of 1135. Across seven hard ProcGen games, our method achieves approximately 3.5x and 5.6x higher aggregate normalized return than the two baselines, respectively.

2026-07-20

RLBRew @ Reinforcement Learning Conference (publié)

To Retain or to Adapt? Generalizing Continual Learning

Giulia Lanzillotta

Mandana Samiei

Razvan Pascanu

Claire Vernade

The Continual Learning (CL) literature has long been driven by the goal of mitigating catastrophic forgetting. This objective rests on a per… (voir plus)vasive, often unstated assumption: that a lifelong learner should approximate the Joint-Task Learning (JTL) solution and retain all previously acquired knowledge. We challenge this retention-centered premise, arguing that in non-stationary environments prioritizing retention can impede real-time adaptation. Shifting the focus to the Average Lifelong Error (ALE), we formalize CL as an online optimization problem governed by the interaction between environmental and learning dynamics. We introduce Transfer Efficiency as a quantitative measure of the tension between Instability, the bias inherited from conflicting past experience, and Transient Error, the optimization cost of learning new tasks from scratch. Under mild convergence conditions, holding across linear and neural network models, this decomposition yields a Critical Task Duration: a closed-form threshold beyond which historical knowledge transitions from a warm-start advantage to an optimization liability whenever retention induces a positive stationary bias. We validate these theoretical predictions on continual image classification and reinforcement learning benchmarks. Finally, by connecting continual learning to the online learning framework of predictable sequences, we show that JTL is only one instance of a broader family of objectives, and we propose a new general class of continual learning algorithms, which we call Predictive Continual Learning. Predictive CL algorithms optimize expected future performance under an explicit, dynamically updated model of future tasks. As a proof of concept, we analyze a Window algorithm that interpolates between JTL and Independent-Task Learning (ITL), outperforming both under controlled distributional drift.

2026-07-05

arXiv (prépublication)

Hierarchical Integration of Predictive Representations of State from General Value Functions

Sonny Jones

Patrick M. Pilarski

Ashley N Dalrymple

In this work, we investigate how predictive representations of state in the form of continually learned General Value Functions (GVFs) inter… (voir plus)act with downstream policy networks. Intelligent agents deployed in real-world environments need to adapt to changing conditions in their environment. Adapting to one’s environment requires a model or representation of the environment on which to base decision-making. Models that take the form of predictions and GVFs have been shown to provide temporally abstracted predictive representations of state that can forecast useful elements of an agent's or environment's future behaviour. While GVFs have been concretely deployed in rehabilitation and robotic domains, existing approaches treat predictions as input features into model frameworks, without examining or comparing how best to integrate them into downstream learning processes. In this work, we compare multiple strategies for integrating observations and GVF predictions into another learning architecture: 1) actual observations solely in the input layer, 2) predictions solely in the input layer, 3) actual observations and predictions in the input layer, and 4) actual observations in the input layer and predictions in the later latent representations. We evaluate these strategies in a rehabilitation setting, using GVFs to learn predictive representations of kinetic and kinematic signals collected from wearable sensors on the lower limb during ambulation across varied terrains, and policy networks to classify walking terrain.

2026-06-09

rl-conference.cc/RLC/2026/Workshop/RL_in_Big_Worlds (poster)

Using Reward Uncertainty to Induce Diverse Behaviour in Reinforcement Learning

Anthony GX-Chen

Ankit Anand

Gheorghe Comanici

Zaheer Abbas

Eser Aygün

David Smalling

Shibl Mourad

Andre Barreto

Mark Rowland

Classical reinforcement learning (RL) typically seeks a deterministic policy that maximizes the expected sum of a scalar reward. Yet, modern… (voir plus) applications such as language model fine-tuning or scientific discovery demand diversity. Existing remedies such as entropy regularization or diversity bonuses often require fragile trade-offs that sacrifice performance for stochasticity or rely on heuristic metrics that can misalign policy rankings. We argue that diversity is more naturally understood as the rational response to uncertainty in the reward. When the reward function is not perfectly known--as is the case with ambiguous preferences or imperfect reward models--committing to a single action can be sub-optimal. Building on this, we propose a fundamental reformulation of the RL objective by replacing the scalar reward with a distribution over reward functions, and applying a non-linear objective over sets of actions. The result is a framework in which calibrated behavioural diversity emerges naturally, remains controllable through the reward function distribution, and is obtained without sacrificing expected reward. Focusing on the contextual bandit setting, we derive a principled gradient estimator for this objective and prove that our formulation naturally generalizes both vanilla policy gradient and more recently developed action-set approaches. Our empirical results demonstrate that this framework offers a robust and theoretically grounded alternative for complex RL tasks where the traditional formulation of the problem fails to induce the desired breadth of agent behaviour.

2026-06-01

arXiv (prépublication)

The schema spectrum: Emergent structures and levels of abstraction in AI and the brain

Mandana Samiei

Blake A. Richards

2026-05-31

Neuron (publié)

Reinforcement Learning with Pairwise Preferences in Long-Term Decision Problems

Jonathan Colaço Carr

Prakash Panangaden

Benjamin Van Roy

Reinforcement learning problems typically define the goal as maximizing the expected value of a scalar reward function. But, pairwise prefer… (voir plus)ences are often easier to specify than scalar rewards, and they express certain goals that scalar rewards cannot. Methods for reinforcement learning with pairwise preferences have thus received growing interest. Unfortunately, these methods are inefficient in problems with long time horizons, and they lack guarantees on the performance of Markov policies relative to history-dependent policies, which bridge the theory and practice of reinforcement learning. We therefore propose the \textit{Markov decision contest} as a new problem model for reinforcement learning with pairwise preferences. We prove that stationary Markov policies are optimal among all history-dependent policies, that solving a Markov decision contest exactly is in P, and that a simple iterative algorithm converges to an optimal policy at a sublinear rate. Lastly, in a set of high-dimensional decision problems with long time horizons, we show that our approximate algorithm is significantly more learning-efficient than prior work.

2026-05-28

arXiv (prépublication)

A systematic review of human-LLM interactions in computational thinking empirical studies

Yimei Zhang

You Song

Reihaneh Rabbany

Maria Cutumisu

2026-05-11

Computer Science Education (publié)

Rotation-Preserving Supervised Fine-Tuning

Mohammad Hamdaqa

Supervised fine-tuning (SFT) improves in-domain performance but can degrade out-of-domain (OOD) generalization. Prior work suggests that thi… (voir plus)s degradation is related to changes in dominant singular subspaces of pretrained weight matrices. However, directly identifying loss-sensitive directions with Hessian or Fisher information is computationally expensive at LLM scale. In this work, we propose preserving projected rotations in pretrained singular subspaces as an efficient proxy for Fisher-sensitive directions, which we call Rotation-Preserving Supervised Fine-Tuning (RPSFT). RPSFT penalizes changes in the projected top-

2026-05-07

arXiv (prépublication)

Balancing Plasticity and Stability with Fast and Slow Successor Features

Raymond Chua

Blake Richards

A hallmark of intelligence is the ability to adapt in non-stationary environments, yet deep Reinforcement Learning (RL) agents often struggl… (voir plus)e in such settings. Prior studies introduce non-stationarity through abrupt shifts in features or dynamics, whereas real-world environments often evolve gradually through continual drift. This distinction has important implications for the ``stability-plasticity dilemma'' in RL, as abrupt task changes may demand more plasticity than naturalistic settings. To address this, we modify existing 3D Miniworld and MuJoCo environments to incorporate naturalistic, continual non-stationarity, and use them to examine how stability and adaptation affect performance under continuous environmental change. We find that methods favoring stability, such as synaptic consolidation, outperform approaches focused on plasticity, such as parameters resetting. Motivated by this result, and prior evidence that Successor Features (SFs) reduce interference, we investigate whether SFs are better consolidation targets than Q-values. Across both environments, applying neuro-inspired synaptic consolidation to SFs yields superior performance on continually changing settings. Moreover, consolidation is most effective when SFs are stabilized across multiple timescales, which capture complementary aspects of gradual environmental change. Together, these results suggest that stability is more critical in continual learning when changes are gradual, and that multi-timescale consolidation of predictive representations is an effective approach.

2026-04-29

International Conference on Machine Learning (accepté)

Nazanin Mohammadi Sepahvand

Detoxifying LLMs via Representation Erasure-Based Preference Optimization

Eleni Triantafillou

Hugo Larochelle

Gintare Karolina Dziugaite

Daniel M. Roy

Large language models (LLMs) trained on webscale data can produce toxic outputs, raising concerns for safe deployment. Prior defenses, based… (voir plus) on applications of DPO, NPO, and similar algorithms, reduce the likelihood of harmful continuations, but not robustly so: they are vulnerable to adversarial prompting and easily undone by fine-tuning-based relearning attacks. Indeed, research has shown that these edits to the model are superficial: linear probing reveals that harmful "directions" remain present in representations. To address this, we propose Representation Erasure-based Preference Optimization (REPO), reformulating detoxification as a token-level preference problem. Using a novel objective with preference data, we force the representations of toxic continuations to converge toward their benign counterparts. Our mechanistic analysis reveals that this granular approach is critical: unlike baselines, REPO induces deep, localized edits to toxicity-encoding neurons while preserving general model utility. Exhaustive evaluations show that REPO achieves state-of-the-art robustness, stopping sophisticated threats-including relearning attacks and enhanced GCG jailbreaks-where existing representation- and output-based methods fail.

2026-02-23

arXiv (inconnu)

Fluid-Agent Reinforcement Learning

Shishir Sharma

Theodore J. Perkins

The primary focus of multi-agent reinforcement learning (MARL) has been to study interactions among a fixed number of agents embedded in an … (voir plus)environment. However, in the real world, the number of agents is neither fixed nor known a priori. Moreover, an agent can decide to create other agents (for example, a cell may divide, or a company may spin off a division). In this paper, we propose a framework that allows agents to create other agents; we call this a fluid-agent environment. We present game-theoretic solution concepts for fluid-agent games and empirically evaluate the performance of several MARL algorithms within this framework. Our experiments include fluid variants of established benchmarks such as Predator-Prey and Level-Based Foraging, where agents can dynamically spawn, as well as a new environment we introduce that highlights how fluidity can unlock novel solution strategies beyond those observed in fixed-population settings. We demonstrate that this framework yields agent teams that adjust their size dynamically to match environmental demands.

2026-02-15

ArXiv (prépublication)