Publications

Implementing a Domain-Adapted GPT Model to Support Genetic Counseling in a Genodermatology Clinic
Megha Udupa
Lorena Alexandra Mija
Ethan Bendayan
Ahmad Alamari
Maxine Joly-Chevrier
Josianne Leblanc
J.-C. Tardif
Elvis Martinez-Jaramillo
Hacene Nedjar
Elham Rahme
William Foulkes
Tania Cruz Marino
Elena Netchiporouk
A novel Al4SiC4–xNx solid solution: synthesis, microstructure, and properties via combined theoretical and experimental approaches
Xu Cheng
Guangchao Xing
Chengji Deng
Yufeng Chen
Zhenglong Liu
Beiyue Ma
Hongxi Zhu
Chao Yu
Resting-state neural oscillations predict individual differences in verbal learning and encoding strategy use
Victor Oswald
Mathieu Landry
Sarah Lippé
Philippe Robaey
Individuals adopt different encoding strategies to facilitate learning, yet few studies have examined the neurophysiological basis of these … (see more)strategies across individuals. The present work addresses this gap by extending our previous findings on the direct relationship between cortical spectral power, measured via resting-state magnetoencephalography, and standard cognitive performance, to test whether resting-state neural features predict individual differences in encoding strategy preferences. Our results highlight the complex interactions between endogenous brain oscillations, learning, and verbal encoding strategies assessed by the California Verbal Learning Test-Second Edition (CVLT-2). First, resting-state theta oscillations were significantly associated with verbal learning and subjective clustering strategies. Second, semantic clustering was facilitated by oscillatory patterns in the left sensory-motor regions. Finally, serial and semantic clustering strategies showed opposite regression patterns, indicating a competitive interaction. Together, these findings provide insights into resting-state neural markers associated with diverse encoding strategies in verbal learning.
SHINIER: An open-source Python package for controlling low-level image properties
Mathias Salvas-Hébert
Nicolas Dupuis-Roy
Catherine Landry
Frédéric Gosselin
CURATE: Automatic Curriculum Learning for Reinforcement Learning Agents through Competence-Based Curriculum Policy Search in Structured Task Spaces
Nan Rosemary Ke
Sarvesh Patil
Annya Dahmani
Eunice Yiu
Alison Gopnik
Oliver Kroemer
Due to fundamental exploration challenges without informed priors or specialized algorithms, agents may be unable to consistently receive in… (see more)formative rewards, leading to inefficient or intractable learning. To address these challenges, we introduce CURATE, an automatic curriculum learning algorithm for reinforcement learning agents in structured task spaces of monotonic difficulty. Through "exploration by exploitation," CURATE dynamically scales the task difficulty to match the agent's current competence. By exploiting its current capabilities that were learned in easier tasks, the agent improves its exploration in more difficult tasks. Our key insight is that the learning improvement in tasks that are close to those used for training is inversely proportional to their difficulty, and an agent that chooses a nearby distribution of the easiest unsolved tasks at any given time can automatically induce an easiest-to-hardest curriculum in these task spaces. To achieve this, CURATE conducts policy search in the task space to learn the best task distribution for training. As the agent's mastery grows, the learned curriculum adapts in an approximately easiest-to-hardest and task-directed fashion, efficiently culminating in a performant agent. Our experiments across three diverse domains (MiniGrid, Procgen, BipedalWalker) demonstrate that CURATE learns effective curricula for sample efficiency and proficiency with the potential for yielding broadly capable agents, matching or exceeding prior curriculum methods that do not require informed initialization or predefined schedules.
One Policy Is Enough: Single-Agent Reinforcement Learning Outperforms Tree Search for Chemistry Tool Learning
Chemistry questions often demand exact computation and database lookups that a language model cannot supply from its parameters, so it must … (see more)reach for external tools. Tool use here is a three-part problem: select the right tool from a large pool, fill it with correctly typed arguments, and chain calls so that each consumes the outputs of the last. CheMatAgent, a previously published system, addresses this with hierarchical evolutionary MCTS: separate policy and execution models searching tool-call trees under two learned critics, one regressed partly onto GPT-assigned scores. We show that a single policy suffices. Our model interleaves reasoning, tool calls, and returns in one left-to-right generation, trained by a supervised warm-up and then outcome-level reinforcement learning against a programmatic reward read directly off the gold call chain, which leaves no learned critic and no judge in the training loop. On ChemToolBench multiple-tool comprehensive chemistry, on both backbones CheMatAgent use, we improve Tool F1 by 5.5% and Return F1 by 9.6% on Qwen-2.5-7B, and by 3.7% and 3.9% on Llama-3.1-8B, compared with their strongest search configuration, at one model invocation per question, against a search whose cost grows with the tree; we also lead answer Pass Rate on Qwen-2.5-7B.
Rare and Common Germline and Somatic Variants Shape Immune Cytopenia Risk and Enable Risk Stratification
Stennio Da Silva Faria
Julie Bineau
Rafael Moisan
Marc‐André Legault
Estelle Lecluze
Thomas Pincez
Abstract The genetic risk factors of immune cytopenias are unclear. Immune cytopenias have been reported in various genetic contexts: 1) inh… (see more)erited error of immunity genes, mainly due to rare germline variants, 2) systemic lupus erythematosus, associated with common germline variants, 3) hematological malignancies, and 4) clonal hematopoiesis, the latter two due to somatic variants. However, the respective contribution and interaction of these variants remain to be investigated. Here, we used two large biobanks with whole genome sequencing data to systematically investigate the genetic contribution to immune cytopenia. We found that the four types of genetic variants independently contribute to immune cytopenia risk. We notably found that carriers of variants in some autosomal recessive genes of inherited error of immunity had an increased risk of immune cytopenia. Additionally, common variant- mediated risk of systemic lupus erythematosus also increased the risk of immune cytopenia. Overall, a third to a half of patients with immune cytopenia carried at least one of the four genetic risk variants investigated. Combining the four variants allowed stratifying the risk of immune cytopenia in both general and high-risk population. In general population, the 10- year incidence of immune cytopenia in the lowest and highest risk groups was 0.8‰ and 15‰, respectively. In sum, this work identified that different genetic risk factors can lead to immune cytopenia. A large proportion of individuals with immune cytopenia carried an underlying genetic risk factor. Finally, combining these genetic risk factors enabled risk stratification.
Revisiting Feedback-Driven LLM Code Repair: A Replication and Exploratory Java Extension
Wassim Keddache
Thomas Perron Touchette
Leuson Da Silva
Since the advent of Large Language Models (LLMs), practitioners have increasingly leveraged them to support their software engineering tasks… (see more), including automated code repair, showing promising results. Yet, concerns regarding reproducibility and generalizability remain largely unexplored. To further evaluate these concerns and associated impacts, we partially reproduce and conduct an exploratory Java extension of the FeedbackEval benchmark [1], which evaluates how LLMs leverage different feedback types for Python code repair. First, we partially replicate the original study on 394 repair tasks using GPT-4o and Claude 3.5 Sonnet, reproducing and observing the main qualitative trends reported in the original work. Second, we conduct an exploratory Java extension by constructing 100 erroneous repair instances from 50 Java tasks and evaluating feedback effectiveness. Our results show that previous conclusions from Python may be sensitive to benchmark construction, feedback representation, and tooling ecosystem, motivating more controlled multilingual benchmarks. Specifically, while test feedback remains the strongest feedback type in our Python replication, the same ranking is not observed in our Java extension, as simple and JUnit-based test feedback do not differ significantly. We hypothesize that differences in feedback informativeness and tooling ecosystems, such as the verbosity of test frameworks, may partly explain such a difference. Finally, lighter prompts reduce cost without significant differences in repair effectiveness. Overall, our findings confirm key trends under a partially controlled replication and highlight the need for more rigorous multilingual evaluation and careful feedback design in LLMbased repair systems.
Simulation-free Structure Learning for Stochastic Population Dynamics
Modeling dynamical systems and unraveling their underlying structural dependencies is central to many domains in the natural sciences. Vario… (see more)us physical systems, such as those arising in cell biology, are inherently high-dimensional and stochastic in nature, and admit only partial, noisy state measurements. Our primary motivating setting is single-cell biology, where destructive measurements yield unpaired population snapshots rather than longitudinal trajectories of the same cells. This poses a significant challenge for addressing the problems of modeling the underlying dynamics and inferring the network structure of these systems. Existing methods are typically tailored either for structure learning or modeling dynamics at the population level, but are limited in their ability to address both problems together. In this work, we address both problems simultaneously: we present StructureFlow, a novel and principled simulation-free training approach for jointly learning the structure and stochastic population dynamics of physical systems. We showcase the utility of StructureFlow for the tasks of structure learning from interventions and dynamical (trajectory) inference of conditional population dynamics. We empirically evaluate our approach on high-dimensional synthetic systems, a set of biologically plausible simulated systems, and an experimental single-cell dataset. We show that StructureFlow can learn the structure of underlying systems while simultaneously modeling their conditional population dynamics --- a key step toward model-based mechanistic understanding of systems behavior.
Temporal coupling as a design principle for social AI
Annemarie Wolff
Vincent Chamberland
Despite their ability to emulate social reasoning, current AI systems consistently fail at the dynamic, real-time coordination that defines … (see more)human social interaction. We argue this failure reflects a missing aspect in their design, not a limitation of scale or training data. Current AI systems model their partner from the outside and train on the data from interactions. Social intelligence, however, is not a property of an individual. It emerges when two coupled individuals interact and is developed through that coupling. We here argue that social AI should be built, and evaluated, as a coupled system with a human which targets an optimal, rather than maximal, alignment. This argument is based on two principles from research in social neuroscience. First, the training objective must change from predicting their interacting partner to a directed coupling that optimizes synchrony. Second, optimizing synchrony requires an embodied and organized temporal architecture which separates faster interactive generation with slower contextual comprehension. We show how coupling can be measured against benchmarks of human interaction and argue that this framing is a safety requirement. Together, these principles offer a neurobiologically grounded pathway toward AI systems capable of real-time social coordination with human partners.
TPR-Attention for Combinatorial Generalization
Systematic generalization remains a significant challenge in deep learning. In particular, combinatorial generalization - generalizing to ne… (see more)w configurations of known factors of variation - is effortless for humans but difficult for standard neural architectures that rely on statistical correlations rather than explicit structural representations. We introduce a new architectural component that embeds structured inductive bias into deep learning: an attention mechanism operating over tensor-product representations (TPRs). Through controlled experiments on compositional tasks, we show that this TPR-attention mechanism outperforms existing architectural components in combinatorial generalization. These results highlight the value of integrating explicit compositional structure into neural attention and point toward a promising path for models capable of systematic generalization.
Beyond Document Grounding: Span-Level Hallucination Detection over Code, Tool Output, and Documents
Ádám Kovács
Bowei He
Xue Liu
István Boros
Szilveszter Tóth
Gábor Recski
Hallucination detection for retrieval-augmented generation (RAG) is usually evaluated on natural-language document evidence. However, ground… (see more)ed generation systems increasingly rely on structured inputs: source code, developer-tool output, markdown documents, tables, and repository metadata. We introduce a unified benchmark for span-level hallucination detection over code, tool output, structured documents, and existing natural-language RAG datasets. The benchmark is built by starting from grounded correct answers, injecting localized hallucinations with exact character labels, and validating the code test split with evidence-based review. Our fine-tuned Qwen3.5-2B detector reaches 0.689 span-F1 on the unified test set and 0.60 on the code-agent source, where it substantially outperforms LettuceDetect-large (0.17) and the strongest zero-shot LLM judges we evaluated (at most 0.22). The same model remains competitive on established natural-language benchmarks, with 81.8 RAGTruth example-F1 and 0.724 English PsiloQA IoU.