Publications

TrackPGD: Efficient Adversarial Attack using Object Binary Masks against Robust Transformer Trackers
Fatemeh Nourilenjan Nokabadi
Yann Batiste Pequignot
Jean-François Lalonde
Christian Gagné
Adversarial perturbations can deceive neural networks by adding small, imperceptible noise to the input. Recent object trackers with transfo… (voir plus)rmer backbones have shown strong performance on tracking datasets, but their adversarial robustness has not been thoroughly evaluated. While transformer trackers are resilient to black-box attacks, existing white-box adversarial attacks are not universally applicable against these new transformer trackers due to differences in backbone architecture. In this work, we introduce TrackPGD, a novel white-box attack that utilizes predicted object binary masks to target robust transformer trackers. Built upon the powerful segmentation attack SegPGD, our proposed TrackPGD effectively influences the decisions of transformer-based trackers. Our method addresses two primary challenges in adapting a segmentation attack for trackers: limited class numbers and extreme pixel class imbalance. TrackPGD uses the same number of iterations as other attack methods for tracker networks and produces competitive adversarial examples that mislead transformer and non-transformer trackers such as MixFormerM, OSTrackSTS, TransT-SEG, and RTS on datasets including VOT2022STS, DAVIS2016, UAV123, and GOT-10k.
2025-05-26
Proceedings of the Conference on Robots and Vision (publié)
doi.org
openreview.net
Adaptive Inertia Estimation of a Power System Area Using Variable Forgetting Factor Recursive Least-Squares
Anoosh Dini
Keyhan Sheshyekani
Hanane Dagdougui
The increasing use of converter-interfaced generators (CIGs) in modern power grids has affected system inertia and posed challenges to grid … (voir plus)stability. In this regard, accurate and real-time monitoring of inertia is crucial for maintaining system stability, especially in low-inertia grids where even small disturbances can lead to rapid frequency deviations. This paper proposes a novel approach for inertia estimation using a variable forgetting factor recursive least squares (VFF-RLS) algorithm, which dynamically adapts to time-varying conditions in power systems. By using ambient measurements provided by the widearea measurement system (WAMS), the proposed approach can capture inertia variations of areas in power systems. The method is validated through simulations on the IEEE 39-bus system, demonstrating higher accuracy compared to existing approaches under both time-constant and time-varying inertia conditions.
2025-05-25
Canadian Conference on Electrical and Computer Engineering (publié)
doi.org
ALAS: Measuring Latent Speech-Text Alignment For Spoken Language Understanding In Multimodal LLMs
Pooneh Mousavi
Yingzhi Wang
Mirco Ravanaelli
Yusuf Cem Sübakan
2025-05-25
ArXiv (prépublication)
doi.org
arxiv.org
Disentangled Source-Free Personalization for Facial Expression Recognition with Neutral Target Data
Masoumeh Sharafi
Emma Ollivier
Muhammad Osama Zeeshan
Soufiane Belharbi
Marco Pedersoli
Alessandro Lameiras Koerich
Simon Bacon
Eric Granger
2025-05-25
2025 IEEE 19th International Conference on Automatic Face and Gesture Recognition (FG) (publié)
doi.org
arxiv.org
Improving Multilingual Math Reasoning for African Languages
Odunayo Ogundepo
Akintunde Oladipo
Kelechi Ogueji
Esther Adenuga
David Ifeoluwa Adelani
Jimmy Lin
Researchers working on low-resource languages face persistent challenges due to limited data availability and restricted access to computati… (voir plus)onal resources. Although most large language models (LLMs) are predominantly trained in high-resource languages, adapting them to low-resource contexts, particularly African languages, requires specialized techniques. Several strategies have emerged for adapting models to low-resource languages in todays LLM landscape, defined by multi-stage pre-training and post-training paradigms. However, the most effective approaches remain uncertain. This work systematically investigates which adaptation strategies yield the best performance when extending existing LLMs to African languages. We conduct extensive experiments and ablation studies to evaluate different combinations of data types (translated versus synthetically generated), training stages (pre-training versus post-training), and other model adaptation configurations. Our experiments focuses on mathematical reasoning tasks, using the Llama 3.1 model family as our base model.
2025-05-25
ArXiv (prépublication)
doi.org
arxiv.org
SCAR: Shapley Credit Assignment for More Efficient RLHF
Meng Cao
Shuyuan Zhang
Xiao-Wen Chang
Doina Precup
2025-05-25
ArXiv (prépublication)
doi.org
arxiv.org
The NaijaVoices Dataset: Cultivating Large-Scale, High-Quality, Culturally-Rich Speech Data for African Languages
Chris Emezue
The NaijaVoices Community
Busayo Awobade
Abraham Owodunni
Handel Emezue
Gloria Monica Tobechukwu Emezue
N. N. Emezue
Sewade Ogun
Bunmi Akinremi
David Ifeoluwa Adelani
Christopher Pal
2025-05-25
ArXiv (prépublication)
doi.org
arxiv.org
BAH Dataset for Ambivalence/Hesitancy Recognition in Videos for Behavioural Change
Manuela Gonz'alez-Gonz'alez
Soufiane Belharbi
Muhammad Osama Zeeshan
Masoumeh Sharafi
Muhammad Haseeb Aslam
Marco Pedersoli
Alessandro Lameiras Koerich
Simon Bacon
Eric Granger
Recognizing complex emotions linked to ambivalence and hesitancy (A/H) can play a critical role in the personalization and effectiveness of … (voir plus)digital behaviour change interventions. These subtle and conflicting emotions are manifested by a discord between multiple modalities, such as facial and vocal expressions, and body language. Although experts can be trained to identify A/H, integrating them into digital interventions is costly and less effective. Automatic learning systems provide a cost-effective alternative that can adapt to individual users, and operate seamlessly within real-time, and resource-limited environments. However, there are currently no datasets available for the design of ML models to recognize A/H. This paper introduces a first Behavioural Ambivalence/Hesitancy (BAH) dataset collected for subject-based multimodal recognition of A/H in videos. It contains videos from 224 participants captured across 9 provinces in Canada, with different age, and ethnicity. Through our web platform, we recruited participants to answer 7 questions, some of which were designed to elicit A/H while recording themselves via webcam with microphone. BAH amounts to 1,118 videos for a total duration of 8.26 hours with 1.5 hours of A/H. Our behavioural team annotated timestamp segments to indicate where A/H occurs, and provide frame- and video-level annotations with the A/H cues. Video transcripts and their timestamps are also included, along with cropped and aligned faces in each frame, and a variety of participants meta-data. We include results baselines for BAH at frame- and video-level recognition in multi-modal setups, in addition to zero-shot prediction, and for personalization using unsupervised domain adaptation. The limited performance of baseline models highlights the challenges of recognizing A/H in real-world videos. The data, code, and pretrained weights are available.
2025-05-24
ArXiv (prépublication)
doi.org
arxiv.org
Response letter to “Confounding by indication and exposure misclassification may undermine corticosteroid effect estimates in ICU patients with alcohol-related hepatitis”
Guillaume Dumas
Maxime Gasperment
Hafid AIT-OUFELLA
2025-05-23
Annals of Intensive Care (publié)
doi.org
Introduction to the special issue on Computational Terminology
Ayla Rigouts Terryn
Patrick Drouin
2025-05-22
Terminology. International Journal of Theoretical and Applied Issues in Specialized Communication (publié)
doi.org
Structure-Aligned Protein Language Model
Can Chen
David Heurtel-Depeiges
Robert M. Vernon
Christopher J. Langmead
Yoshua Bengio
Quentin Fournier
2025-05-21
ArXiv (prépublication)
doi.org
arxiv.org
Tumor antigens preferentially derive from unmutated genomic sequences in melanoma and non-small cell lung cancer
Anca Apavaloaei
Qingchuan Zhao
Leslie Hesnard
Maxime Cahuzac
Chantal Durette
Jean-David Larouche
Marie-Pierre Hardy
Krystel Vincent
Sylvie Brochu
Jean-Philippe Laverdure
Joël Lanoix
Mathieu Courcelles
Patrick Gendron
Mathieu Lajoie
Maria Virginia Ruiz Cuevas
Eralda Kina
Julie Perrault
Juliette Humeau
Gregory Ehx
Sébastien Lemieux … (voir 5 de plus)
Ian R. Watson
Daniel E. Speiser
Michal Bassani-Sternberg
Pierre Thibault
Claude Perreault
Melanoma and non-small cell lung cancer (NSCLC) display exceptionally high mutational burdens. Hence, immune targeting in these cancers has … (voir plus)primarily focused on tumor antigens (TAs) predicted to derive from nonsynonymous mutations. Using comprehensive proteogenomic analyses, we identified 589 TAs in cutaneous melanoma (n = 505) and NSCLC (n = 90). Of these, only 1% were derived from mutated sequences, which was explained by a low RNA expression of most nonsynonymous mutations and their localization outside genomic regions proficient for major histocompatibility complex (MHC) class I-associated peptide generation. By contrast, 99% of TAs originated from unmutated genomic sequences specific to cancer (aberrantly expressed tumor-specific antigens (aeTSAs), n = 220), overexpressed in cancer (tumor-associated antigens (TAAs), n = 165) or specific to the cell lineage of origin (lineage-specific antigens (LSAs), n = 198). Expression of aeTSAs was epigenetically regulated, and most were encoded by noncanonical genomic sequences. aeTSAs were shared among tumor samples, were immunogenic and could contribute to the response to immune checkpoint blockade observed in previous studies, supporting their immune targeting across cancers.
2025-05-21
Nature Cancer (publié)
doi.org