Doina Precup

guangyuan.wang@mila.quebec

Guangyuan Wang

Stagiaire de recherche - McGill University

Haque Ishfaq

Doctorat - McGill University

Doctorat - McGill University

huanghow@mila.quebec

Janarthanan Rajendran

Postdoctorat - Université de Montréal

Superviseur⋅e principal⋅e :

Sarath Chandar Anbil Parthipan

janarthanan.rajendran@mila.quebec

jonathan.colaco-carr@mila.quebec

Jaume Minano Masip

Doctorat - McGill University

masipmij@mila.quebec

Jesse Farebrother

Doctorat - McGill University

Superviseur⋅e principal⋅e :

Marc Gendron-Bellemare

Maîtrise recherche - McGill University

Superviseur⋅e principal⋅e :

Prakash Panangaden

Jonathan Lebensold

Doctorat - McGill University

Stagiaire de recherche - McGill University

keyu.wang@mila.quebec

Kushal Arora

Doctorat - McGill University

Superviseur⋅e principal⋅e :

Lynn Cherif

Maîtrise recherche - McGill University

Co-superviseur⋅e :

Khimya Khetarpal

lynn.cherif@mila.quebec

Mohammad Sami Nur Islam Islam

Mandana Samiei

Doctorat - McGill University

Co-superviseur⋅e :

Doctorat - McGill University

delvermm@mila.quebec

Martin Klissarov

Doctorat - McGill University

Harry Zhao

Doctorat - McGill University

Co-superviseur⋅e :

Stagiaire de recherche - McGill University

mohammad-sami-nur.islam@mila.quebec

nathan.de-lara@mila.quebec

Nathan de Lara

Stagiaire de recherche - McGill University

Nate Rahn

Doctorat - McGill University

Superviseur⋅e principal⋅e :

Marc Gendron-Bellemare

nathan.rahn@mila.quebec

Girdhar Neil Girdhar

Collaborateur·rice de recherche - McGill University

neil.girdhar@mila.quebec

Nikhil Vemgal

Maîtrise recherche - McGill University

nikhil-murali.vemgal@mila.quebec

padideh.nouri@mila.quebec

Nishanth Anand Vemgal

Doctorat - McGill University

Maîtrise recherche - Université de Montréal

Doctorat - McGill University

Ray Chua

Doctorat - McGill University

Co-superviseur⋅e :

Blake Richards

chuaraym@mila.quebec

Riashat Islam

Doctorat - McGill University

Safa Alver

Doctorat - McGill University

alversaf@mila.quebec

Sahand Rezaei-Shoshtari

Doctorat - McGill University

Co-superviseur⋅e :

David Meger

sahand.rezaei-shoshtari@mila.quebec

Doctorat - McGill University

Doctorat - McGill University

Co-superviseur⋅e :

David Meger

fujimots@mila.quebec

Shahrad Mohammadzadeh

Collaborateur·rice de recherche - McGill University

Superviseur⋅e principal⋅e :

Reihaneh Rabbany

shahrad.mohammadzadeh@mila.quebec

Doctorat - McGill University

Doctorat - McGill University

shuyuan.zhang@mila.quebec

Sitao Luan

Doctorat - McGill University

Steve Wen

Baccalauréat - McGill University

steve.wen@mila.quebec

Sumana Basu

Doctorat - McGill University

Co-superviseur⋅e :

Adriana Romero Soriano

Maîtrise recherche - Université de Montréal

Superviseur⋅e principal⋅e :

Yoshua Bengio

thomas.jiralerspong@mila.quebec

Doctorat - McGill University

cheluver@mila.quebec

Wesley Chung

Doctorat - McGill University

Superviseur⋅e principal⋅e :

David Meger

chungwes@mila.quebec

Ray Luo

Doctorat - McGill University

Superviseur⋅e principal⋅e :

Xujie Si

luo.ziyan@mila.quebec

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

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

Lire l'article

Publications

The Stable Entropy Hypothesis and Entropy-Aware Decoding: An Analysis and Algorithm for Robust Natural Language Generation

Kushal Arora

Timothy O'Donnell

Jason Aaron Edward Weston

Jackie C.K.Cheung

State-of-the-art language generation models can degenerate when applied to open-ended generation problems such as text completion, story gen… (voir plus)eration, or dialog modeling. This degeneration usually shows up in the form of incoherence, lack of vocabulary diversity, and self-repetition or copying from the context. In this paper, we postulate that ``human-like'' generations usually lie in a narrow and nearly flat entropy band, and violation of these entropy bounds correlates with degenerate behavior. Our experiments show that this stable narrow entropy zone exists across models, tasks, and domains and confirm the hypothesis that violations of this zone correlate with degeneration. We then use this insight to propose an entropy-aware decoding algorithm that respects these entropy bounds resulting in less degenerate, more contextual, and"human-like"language generation in open-ended text generation settings.

2023-02-14

ArXiv (prépublication)

Multitask Reinforcement Learning by Optimizing Neural Pathways

Samin Yeasar Arnob

Riyasat Ohib

Amy Zhang

Sergey Plis

Reinforcement learning (RL) algorithms have achieved great success in learning specific tasks, as evidenced by examples such as AlphaGo or f… (voir plus)usion control. However, it is still difficult for an RL agent to learn how to solve multiple tasks. In this paper, we propose a novel multitask learning framework, in which multiple specialized pathways through a single network are trained simultaneously, with each pathway focusing on a single task. We show that this approach achieves competitive performance with existing multitask RL methods, while using only 5% of the number of neurons per task. We demonstrate empirically the success of our approach on several continuous control tasks, in both online and offline training.

2023-02-01

ICLR.cc/2023/Conference (rejected)

Minimal Value-Equivalent Partial Models for Scalable and Robust Planning in Lifelong Reinforcement Learning

Safa Alver

2023-01-24

ArXiv (preprint)

On the Challenges of using Reinforcement Learning in Precision Drug Dosing: Delay and Prolongedness of Action Effects

Sumana Basu

M. Legault

Adriana Romero Soriano

Drug dosing is an important application of AI, which can be formulated as a Reinforcement Learning (RL) problem. In this paper, we identify … (voir plus)two major challenges of using RL for drug dosing: delayed and prolonged effects of administering medications, which break the Markov assumption of the RL framework. We focus on prolongedness and define PAE-POMDP (Prolonged Action Effect-Partially Observable Markov Decision Process), a subclass of POMDPs in which the Markov assumption does not hold specifically due to prolonged effects of actions. Motivated by the pharmacology literature, we propose a simple and effective approach to converting drug dosing PAE-POMDPs into MDPs, enabling the use of the existing RL algorithms to solve such problems. We validate the proposed approach on a toy task, and a challenging glucose control task, for which we devise a clinically-inspired reward function. Our results demonstrate that: (1) the proposed method to restore the Markov assumption leads to significant improvements over a vanilla baseline; (2) the approach is competitive with recurrent policies which may inherently capture the prolonged affect of actions; (3) it is remarkably more time and memory efficient than the recurrent baseline and hence more suitable for real-time dosing control systems; and (4) it exhibits favourable qualitative behavior in our policy analysis.

2023-01-02

ArXiv (prépublication)

Membership Inference Attacks Against Temporally Correlated Data in Deep Reinforcement Learning

Maziar Gomrokchi

Susan Amin

Hossein Aboutalebi

Alexander Wong

While significant research advances have been made in the field of deep reinforcement learning, there have been no concrete adversarial atta… (voir plus)ck strategies in literature tailored for studying the vulnerability of deep reinforcement learning algorithms to membership inference attacks. In such attacking systems, the adversary targets the set of collected input data on which the deep reinforcement learning algorithm has been trained. To address this gap, we propose an adversarial attack framework designed for testing the vulnerability of a state-of-the-art deep reinforcement learning algorithm to a membership inference attack. In particular, we design a series of experiments to investigate the impact of temporal correlation, which naturally exists in reinforcement learning training data, on the probability of information leakage. Moreover, we compare the performance of collective and individual membership attacks against the deep reinforcement learning algorithm. Experimental results show that the proposed adversarial attack framework is surprisingly effective at inferring data with an accuracy exceeding 84% in individual and 97% in collective modes in three different continuous control Mujoco tasks, which raises serious privacy concerns in this regard. Finally, we show that the learning state of the reinforcement learning algorithm influences the level of privacy breaches significantly.

2023-01-01

IEEE Access (publié)

Multi-Environment Pretraining Enables Transfer to Action Limited Datasets

David Venuto

Sherry Yang

Pieter Abbeel

Igor Mordatch

Ofir Nachum

2023-01-01

ICML (publié)

Multi-Environment Pretraining Enables Transfer to Action Limited Datasets

David Venuto

Sherry Yang

Pieter Abbeel

Igor Mordatch

Ofir Nachum

Using massive datasets to train large-scale models has emerged as a dominant approach for broad generalization in natural language an… (voir plus)d vision applications. In reinforcement learning, however, a key challenge is that available data of sequential decision making is often not annotated with actions - for example, videos of game-play are much more available than sequences of frames paired with the logged game controls. We propose to circumvent this challenge by combining large but sparsely-annotated datasets from a \emph{target} environment of interest with fully-annotated datasets from various other \emph{source} environments. Our method, Action Limited PreTraining (ALPT), leverages the generalization capabilities of inverse dynamics modelling (IDM) to label missing action data in the target environment. We show that utilizing even one additional environment dataset of labelled data during IDM pretraining gives rise to substantial improvements in generating action labels for unannotated sequences. We evaluate our method on benchmark game-playing environments and show that we can significantly improve game performance and generalization capability compared to other approaches, even when using annotated datasets equivalent to only 12 minutes of gameplay.

2023-01-01

ICML (publié)

Optimism and Adaptivity in Policy Optimization

Veronica Chelu

Tom Zahavy

Arthur Guez

Sebastian Flennerhag

2023-01-01

arXiv.org (prépublication)

Temporal Abstraction in Reinforcement Learning with the Successor Representation

Marlos C. Machado

Andre Barreto

Michael Bowling

When Do Graph Neural Networks Help with Node Classification: Investigating the Homophily Principle on Node Distinguishability

Sitao Luan

Chenqing Hua

Minkai Xu

Qincheng Lu

Jiaqi Zhu

Xiao-Wen Chang

Jie Fu

Jure Leskovec

Homophily principle, i.e., nodes with the same labels are more likely to be connected, was believed to be the main reason for the performanc… (voir plus)e superiority of Graph Neural Networks (GNNs) over Neural Networks (NNs) on Node Classiﬁcation (NC) tasks. Recently, people have developed theoretical results arguing that, even though the homophily principle is broken, the advantage of GNNs can still hold as long as nodes from the same class share similar neighborhood patterns [29], which questions the validity of homophily. However, this argument only considers intra-class Node Distinguishability (ND) and ignores inter-class ND, which is insufﬁcient to study the effect of homophily. In this paper, we ﬁrst demonstrate the aforementioned insufﬁciency with examples and argue that an ideal situation for ND is to have smaller intra-class ND than inter-class ND. To formulate this idea and have a better understanding of homophily, we propose Contextual Stochastic Block Model for Homophily (CSBM-H) and deﬁne two metrics, Probabilistic Bayes Error (PBE) and Expected Negative KL-divergence (ENKL), to quantify ND, through which we can also ﬁnd how intra- and inter-class ND inﬂuence ND together. We visualize the results and give detailed analysis. Through experiments, we veriﬁed that the superiority of GNNs is

Towards Continual Reinforcement Learning: A Review and Perspectives

Khimya Khetarpal

Matthew D Riemer

Irina Rish

2022-12-22

Journal of Artificial Intelligence Research (publié)

Low-Rank Representation of Reinforcement Learning Policies

Bogdan Mazoure

Thang Doan

Tianyu Li

We propose a general framework for policy representation for reinforcement learning tasks. This framework involves finding a low-dimensional… (voir plus) embedding of the policy on a reproducing kernel Hilbert space (RKHS). The usage of RKHS based methods allows us to derive strong theoretical guarantees on the expected return of the reconstructed policy. Such guarantees are typically lacking in black-box models, but are very desirable in tasks requiring stability and convergence guarantees. We conduct several experiments on classic RL domains. The results confirm that the policies can be robustly represented in a low-dimensional space while the embedded policy incurs almost no decrease in returns.

2022-10-27

Journal of Artificial Intelligence Research (publié)