Reihaneh Rabbany

hussein.abdallah@mcgill.ca

Biographie

Reihaneh Rabbany est professeure adjointe à l'École d'informatique de l'Université McGill. Elle est membre du corps professoral de Mila – Institut québécois d’intelligence artificielle et titulaire d'une chaire en IA Canada-CIFAR. Elle est également membre du corps enseignant du Centre pour l’étude de la citoyenneté démocratique de McGill. Avant de se joindre à l’Université McGill, elle a été boursière postdoctorale à la School of Computer Science de l'Université Carnegie Mellon. Elle a obtenu un doctorat à l’Université de l’Alberta, au Département d'informatique. Elle dirige le laboratoire de données complexes, dont les recherches se situent à l'intersection de la science des réseaux, de l'exploration des données et de l'apprentissage automatique, et se concentrent sur l'analyse des données interconnectées du monde réel et sur les applications sociales.

Étudiants actuels

Hussein Abdallah

Postdoctorat - McGill

Doctorat - UdeM

Superviseur⋅e principal⋅e :

oumaima.amezgar@mila.quebec

Aurélien Bück-Kaeffer

Maîtrise recherche - McGill

aurelien.buck-kaeffer@mila.quebec

Andrea Cini

Visiteur de recherche indépendant - Mila

Stagiaire de recherche - UdeM

Superviseur⋅e principal⋅e :

mcorriveau303@gmail.com

Johnny Dellas

Doctorat - McGill

john.dellas@mila.quebec

Islam Eldifrawi

Visiteur de recherche indépendant - University of Sherbrooke

sulum821991@gmail.com

sophia.garrel@mila.quebec

Aarash Feizi

Doctorat - McGill

Co-superviseur⋅e :

Adriana Romero Soriano

Collaborateur·rice de recherche - McGill

Nicholas Gaudet

Doctorat - UdeM

Superviseur⋅e principal⋅e :

nicholas.gaudet@mila.quebec

mattia.greiche@mila.quebec

Mattia Greiche

Stagiaire de recherche - McGill

Sabína Gulčíková

Visiteur de recherche indépendant - Brno University of Technology

sabina.gulcikova@mila.quebec

Nazia Hossain

Collaborateur·rice de recherche - McGill University

nazia.hossain@mila.quebec

Shenyang Huang

Collaborateur·rice de recherche - McGill

Co-superviseur⋅e :

Anne Imouza

Doctorat - McGill

Superviseur⋅e principal⋅e :

anne.imouza@mila.quebec

Emma Kondrup

Doctorat - McGill

Co-superviseur⋅e :

Catherine Régis

emma.kondrup@mila.quebec

Andrew Lin

Stagiaire de recherche - McGill University

andrew.lin@mila.quebec

victor.livernoche@mila.quebec

Victor Livernoche

Doctorat - McGill

shahrad.mohammadzadeh@mila.quebec

Sitao Luan

Postdoctorat - McGill

Superviseur⋅e principal⋅e :

Shahrad Mohammadzadeh

Maîtrise recherche - McGill

Co-superviseur⋅e :

Doina Precup

Postdoctorat - UdeM

Superviseur⋅e principal⋅e :

nardin.nakhla@mila.quebec

Farimah Poursafaei

Collaborateur·rice alumni - McGill

farimah.poursafaei@mila.quebec

Doctorat - UdeM

Superviseur⋅e principal⋅e :

meissa.rebbani@mila.quebec

sebastian.sabry@mila.quebec

Sebastian Sabry

Maîtrise recherche - McGill University

dorsaf.sallami@mila.quebec

Dorsaf Sallami

Visiteur de recherche indépendant - McGill

jacob.mila.handle@tianshome.com

Sneheel Sarangi

Maîtrise recherche - McGill

sneheelsarangi@gmail.com

Maîtrise recherche - McGill University

shiyan.wliu@gmail.com

Vidya Sujaya

Maîtrise recherche - McGill

vidya.sujaya@mila.quebec

Jacob-Junqi Tian

Maîtrise recherche - McGill

Collaborateur·rice de recherche - UdeM

Superviseur⋅e principal⋅e :

Guillaume Rabusseau

aaron.tjandra@mila.quebec

Li Wei Wang

Maîtrise recherche - McGill

liweiwang2002@gmail.com

Tong Wu

Baccalauréat - McGill University

Co-superviseur⋅e :

tong.wu@mila.quebec

Brian Yang

Doctorat - McGill

brian.yang@mila.quebec

Maîtrise recherche - McGill

xueli.zhao@mila.quebec

Sveta Zhuk

Maîtrise recherche - UdeM

Superviseur⋅e principal⋅e :

svetlana.zhuk@mila.quebec

Billets de blogue

Un groupe hétéroclite de huit jeunes adultes se tient serré sur un toit, souriant et riant avec la silhouette de la ville en arrière-plan. Deux encarts circulaires mettent en évidence une caméra vintage tenue par l'un des membres du groupe, soulignant l'élément deepfake.

16 décembre 2025

Démasquer les deepfakes grâce à l'IA

par

Victor Livernoche

Reihaneh Rabbany

Lire l'article

Flight-SEIR: Incorporating Flight Data to Improve Epidemiological Modelling and Disease Outbreak Prevention

3 août 2021

Flight-SEIR : incorporer les données de vol pour améliorer la modélisation épidémiologique et la prévention d’éclosions de maladies infectieuses

par

Shenyang Huang

Reihaneh Rabbany

Lire l'article

Publications

A Framework for Context Enrichment and Selective Robustness in LLM Social Agents

Mattia Greiche

Maximilian Puelma Touzel

Researchers use large language models as agents in social simulations, where agents are prompted with observations and background informatio… (voir plus)n intended to shape their actions. Yet LLM agents have been shown to substantially change their behavior under meaning-preserving prompt variations, creating a validation problem for social simulation. One plausible hypothesis is that this brittleness reflects lack of grounding and can be reduced by enriching the context provided to the agents. Whether richer context improves robustness remains unclear, in part because context richness and robustness are rarely defined in ways that support controlled comparisons. We address the first need with controlled context enrichment, a set of prompt substitution rules that systematically enrich distinct types of context: neutral padding, generic social background, descriptive decision grounding, and normative decision grounding. We address the second by introducing selective robustness as a validation criterion: insensitivity to meaning-preserving prompt variation combined with a directional response to decision-relevant changes in the social world. We demonstrate the framework in a Prisoner's Dilemma-like game, showing how it can be used to identify which forms of context enrichment produce more robust agent behavior. We find that these forms of enrichment produce distinct behavioral patterns that vary between the two models tested.

2026-10-04

SocialAgent @ Neural Information Processing Systems (poster)

Ambiguity-Aware Evaluation for Automated Fact-Checking

Sukanya Krishna

Dorsaf Sallami

Maimuna S. Majumder

Finale Doshi-Velez

Automated fact-checking systems are evaluated almost entirely by whether their predicted verdict matches a single gold label, an approach th… (voir plus)at assumes every claim has one stable interpretation and that the available evidence is always sufficient to resolve it. In practice, claims are frequently underspecified in time, place, or meaning, and retrieved evidence is often incomplete or contradictory, so this approach cannot distinguish a system that reasoned well on a genuinely ambiguous input from one that reasoned poorly but matched the label anyway. We introduce a framework that separates ambiguity in the claim and evidence from failures in the system's own reasoning and grounding, and conditions accuracy and response appropriateness on this ambiguity profile. Applied to a retrieval-based fact-checking pipeline across five benchmarks, plus a human pilot study, we find that ambiguity is pervasive and dominated by missing timeframes, that most correct verdicts on several benchmarks are backed by reasoning that would not hold up under scrutiny, and that systems grow more confident, not less, when their evidence conflicts.

2026-09-29

WiML @ Neural Information Processing Systems (publié)

Foreword to the special issue on responsible artificial intelligence: methodologies, implications, and practices

Ebrahim Bagheri

Robin Cohen

Faezeh Ensan

Benjamin C.M. Fung

Sébastien Gambs

2026-08-13

Information Processing & Management (publié)

CrediBench: Building Web-Scale Network Datasets for Information Integrity

James Zhou

Michael M. Bronstein

Shenyang Huang

Online misinformation poses an escalating threat, amplified by the Internet's open nature and increasingly capable LLMs that generate persua… (voir plus)sive yet deceptive content. Existing misinformation detection methods typically focus on either textual content or network structure in isolation, failing to leverage the rich, dynamic interplay between website content and hyperlink relationships that characterizes real-world misinformation ecosystems. We introduce CrediBench: a large-scale data processing pipeline for constructing temporal web graphs that jointly model textual content and hyperlink structure for misinformation detection. Unlike prior work, our approach captures the dynamic evolution of general misinformation domains, including changes in both content and inter-site references over time. Our processed one-month snapshot extracted from the Common Crawl archive in December 2024 contains 45 million nodes and 1 billion edges, representing the largest web graph dataset made publicly available for misinformation research to date. From our experiments on this graph snapshot, we demonstrate the strength of both structural and webpage content signals for learning credibility scores, which measure source reliability. The pipeline and experimentation code are all available here, and the dataset is in this folder.

2026-08-07

ACM SIGKDD Conference on Knowledge Discovery and Data Mining (publié)

StabilityBench: Benchmarking Instability in LLMs

AI Assistants are increasingly deployed in high-stakes settings, such as healthcare or government services. Yet their real-world behavior re… (voir plus)mains poorly understood due to strong context dependence. Current evaluation protocols follow a defense-in-depth paradigm with compounding layers of safeguards, ranging from traditional benchmarks to live or adversarial testing. Such benchmarks remain largely static and single-turn, limiting their ability to capture real-world variability in conversational settings. We propose StabilityBench, a principled, general and model-agnostic benchmark operator that transforms single-turn benchmark queries into multi-turn interaction histories. StabilityBench augments existing benchmarks by injecting realistic user simulations, through demographic proxies or sycophantic baits, while preserving original task intent. We apply StabilityBench to four benchmarks spanning mathematical reasoning, health question-answering and safety, and evaluate nine large language models under these conditions. Our results show that model performance is consistently unstable under these injections, with considerable performance degradations on three out of four benchmarks studied. These highlight important limitations of static evaluations and motivate more realistic evaluation settings. To this end, we propose StabilityBench-Mini: a size-preserving variant of StabilityBench that samples across diversification axes, enabling more realistic evaluation without increasing costs.

2026-07-16

arXiv (prépublication)

ToxiSight: Leveraging Moderator Expertise Through Behavioral Measurement in Gaming Toxicity Annotation

Vicki Chen

Domenico Tullo

Content moderation systems commonly treat human annotators as interchangeable label sources, resolving disagreements through majority voting… (voir plus) or expert arbitration. We present ToxiSight, an annotation platform that reframes this assumption: rather than extracting consensus, the system supports moderator reasoning by treating hesitation, revision, and disagreement as signals revealing where content is genuinely ambiguous and where taxonomic guidelines fail. ToxiSight integrates gaming-specific contextual widgets with behavioral telemetry, capturing the cognitive processes underlying toxicity validation decisions. Through deployment with 10 professional moderators across 60,000 lines of gaming chat, we demonstrate that behavioral patterns expose systematic category failures invisible to traditional inter-annotator metrics. The Controversial category shows 72% revision rates with fast processing times, indicating immediate recognition of definitional breakdown, while Threats (Life-Threatening) exhibits 75% revisions with slow processing, signaling genuine interpretive complexity. Completion rates improved from 60% to 95%, and moderators reported reduced decision stress when permitted to express uncertainty. This case study demonstrates that trustworthy toxicity detection requires annotation systems designed around the irreducible complexity of human judgment, not against it.

2026-06-28

Canadian Conference on Artificial Intelligence (publié)

proceedings.mlr.press

EASE Configuration Facilitates A Reproducible Science of LLM Social Simulations

Sneheel Sarangi

Maximilian Puelma Touzel

Aurélien Bück-Kaeffer

LLMs are increasingly deployed to simulate social interactions, yet many of the existing simulators remain ad hoc and monolithic. This lack … (voir plus)of architectural standardization prevents reproducible research and complicates downstream evaluation. We advance a rigorous science of LLM-based multi-agent simulation by modularizing core components into Environments, Agents, Simulation engines, and Evaluation metrics (EASE). We demonstrate the utility of EASE configuration by wrapping it in an experimental study schema for orchestrating workflows centered around answering explicit research questions in generated scenarios. We contribute SiliSocS, an open-source, research-ready Silicon Society Sandbox implementing a study-structured EASE configuration to enable highly configurable and reproducible LLM-based social simulations. Using SiliSocS and EASE, we present three case studies, showcasing the system's comprehensive assessment of existing questions, ability to dive deeper into complex questions, and elaboration of existing studies, respectively. Together, these case studies highlight the limitations of current modeling approaches and isolate the impacts of design choices on key results.

2026-05-27

arXiv (prépublication)

Hydra: Towards Transferable Multi-Task Learning on Temporal Graphs

Kiarash Shamsi

Tran Gia Bao Ngo

Baris Coskunuzer

Michael M. Bronstein

Cuneyt Gurcan Akcora

Real-world evolving networks are naturally modeled as temporal graphs (TGs), where capturing temporal dynamics is essential for predicting f… (voir plus)uture graph properties that support downstream decision-making. Existing temporal graph methods have been developed primarily for single-task prediction, and little is known about their generalization across tasks or transfer to unseen networks. This leaves the challenge of multi-task graph property prediction in TGs largely open. We address this challenge by introducing Hydra, a novel architecture that integrates local connectivity features from temporal GNNs with a spectral learning module that captures global connectivity patterns. This design enables joint learning of local and global information under a multi-task objective. In multi-task classification, Hydra achieves an 8.9% relative gain in AUC over the strongest competitor. In multi-task regression, Hydra achieves competitive results in all three tasks, while obtaining the best results in two tasks with a 8.2% relative gain in MAE compared to the strongest baseline. Moreover, Hydra delivers these gains with a 22× reduction in training time compared to temporal transfer models. These results provide the first systematic evidence that multi-task transferable learning on temporal graphs is effective. By delivering consistent top-ranked performance, Hydra highlights multi-task training on temporal graphs as a promising direction toward adaptable foundation models for temporal graphs.

2026-05-20

GFM @ International Conference on Machine Learning (poster)

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

Yimei Zhang

You Song

Doina Precup

Maria Cutumisu

2026-05-11

Computer Science Education (publié)

Kurtosis-Guided Denoising Score Matching for Tabular Anomaly Detection

Victor Livernoche

Jie Zan

Denoising score matching (DSM) provides a way to learn data distributions by training a neural network to recover the score function, define… (voir plus)d as the gradient of the log density, from noise-corrupted samples. Once trained, the score magnitude at a test point reflects how consistent that point is with the learned distribution, making it a natural anomaly signal. The key practical challenge is selecting the perturbation scale: too little noise yields unstable score estimates in sparse regions, while too much erases local structure and weakens anomaly sensitivity. This is compounded by the difficulty of hyperparameter tuning when anomalies are unknown and no validation set is available. We introduce kurtosis-based noise scaling (K-DSM), a per-feature scheme that sets noise levels from the shape of each marginal distribution, improving coverage of low-density regions and precision in high-density regions without extra model complexity. Contrary to prior claims that multi-scale or noise-conditioned training is necessary, we find that a carefully trained single-scale model is already a strong anomaly detector. On standard tabular anomaly detection benchmarks, K-DSM achieves state-of-the-art performance in the semi-supervised setting. When combined with a lightweight EMA-teacher filtering rule that removes low-density training points before each gradient step, it also achieves strong performance in the fully unsupervised (contaminated) setting, suggesting that simple, data-adaptive noise scaling enables robust anomaly detection while reducing reliance on hyperparameter tuning.

2026-05-06

arXiv (prépublication)

The $\textit{Silicon Society}$ Cookbook: Design Space of LLM-based Social Simulations

Aurélien Bück-Kaeffer

Sneheel Sarangi

Maximilian Puelma Touzel

Studies attempting to simulate human behavior with …

2026-04-29

arXiv (prépublication)

Deepfakes in the 2025 Canadian Election: Prevalence, Partisanship, and Platform Dynamics

Victor Livernoche

Andreea Musulan

Concerns about AI-generated political content are growing, yet there is limited empirical evidence on how deepfakes actually appear and circ… (voir plus)ulate across social platforms during major events in democratic countries. In this study, we present one of the first in-depth analyses of how these realistic synthetic media shape the political landscape online, focusing specifically on the 2025 Canadian federal election. By analyzing 187,778 posts from X, Bluesky, and Reddit with a high-accuracy detection framework trained on a diverse set of modern generative models, we find that 5.86% of election-related images were deepfakes. Right-leaning accounts shared them more frequently, with 8.66% of their posted images flagged compared to 4.42% for left-leaning users, often with defamatory or conspiratorial intent. Yet, most detected deepfakes were benign or non-political, and harmful ones drew little attention, accounting for only 0.12% of all views on X. Overall, deepfakes were present in the election conversation, but their reach was modest, and realistic fabricated images, although less common, drew higher engagement, highlighting growing concerns about their potential misuse.

2026-04-11

ACM Web Conference (publié)