Publications

Asymmetric developmental bifurcations in polarized environments: a new class of human variants, which may include autism.
Laurent Mottron
Alix Lavigne-Champagne
Boris C. Bernhardt
Guillaume Dumas
Sébastien Jacquemont
D. Gagnon
2025-09-29
Molecular Psychiatry (publié)
doi.org
BloomAPR: A Bloom's Taxonomy-based Framework for Assessing the Capabilities of LLM-Powered APR Solutions
Yinghang Ma
Jiho Shin
Leuson Da Silva
Zhen Ming (Jack) Jiang
Song Wang
Foutse Khomh
Shin Hwei Tan
Recent advances in large language models (LLMs) have accelerated the development of AI-driven automated program repair (APR) solutions. Howe… (voir plus)ver, these solutions are typically evaluated using static benchmarks such as Defects4J and SWE-bench, which suffer from two key limitations: (1) the risk of data contamination, potentially inflating evaluation results due to overlap with LLM training data, and (2) limited ability to assess the APR capabilities in dynamic and diverse contexts. In this paper, we introduced BloomAPR, a novel dynamic evaluation framework grounded in Bloom's Taxonomy. Our framework offers a structured approach to assess the cognitive capabilities of LLM-powered APR solutions across progressively complex reasoning levels. Using Defects4J as a case study, we evaluated two state-of-the-art LLM-powered APR solutions, ChatRepair and CigaR, under three different LLMs: GPT-3.5-Turbo, Llama-3.1, and StarCoder-2. Our findings show that while these solutions exhibit basic reasoning skills and effectively memorize bug-fixing patterns (fixing up to 81.57% of bugs at the Remember layer), their performance increases with synthetically generated bugs (up to 60.66% increase at the Understand layer). However, they perform worse on minor syntactic changes (fixing up to 43.32% at the Apply layer), and they struggle to repair similar bugs when injected into real-world projects (solving only 13.46% to 41.34% bugs at the Analyze layer). These results underscore the urgent need for evolving benchmarks and provide a foundation for more trustworthy evaluation of LLM-powered software engineering solutions.
2025-09-29
ArXiv (prépublication)
arxiv.org
arxiv.org
ClustRecNet: A Novel End-to-End Deep Learning Framework for Clustering Algorithm Recommendation
Mohammadreza Bakhtyari
Bogdan Mazoure
Renato Cordeiro De Amorim
Guillaume Rabusseau
Vladimir Makarenkov
We introduce ClustRecNet - a novel deep learning (DL)-based recommendation framework for determining the most suitable clustering algorithms… (voir plus) for a given dataset, addressing the long-standing challenge of clustering algorithm selection in unsupervised learning. To enable supervised learning in this context, we construct a comprehensive data repository comprising 34,000 synthetic datasets with diverse structural properties. Each of them was processed using 10 popular clustering algorithms. The resulting clusterings were assessed via the Adjusted Rand Index (ARI) to establish ground truth labels, used for training and evaluation of our DL model. The proposed network architecture integrates convolutional, residual, and attention mechanisms to capture both local and global structural patterns from the input data. This design supports end-to-end training to learn compact representations of datasets and enables direct recommendation of the most suitable clustering algorithm, reducing reliance on handcrafted meta-features and traditional Cluster Validity Indices (CVIs). Comprehensive experiments across synthetic and real-world benchmarks demonstrate that our DL model consistently outperforms conventional CVIs (e.g. Silhouette, Calinski-Harabasz, Davies-Bouldin, and Dunn) as well as state-of-the-art AutoML clustering recommendation approaches (e.g. ML2DAC, AutoCluster, and AutoML4Clust). Notably, the proposed model achieves a 0.497 ARI improvement over the Calinski-Harabasz index on synthetic data and a 15.3% ARI gain over the best-performing AutoML approach on real-world data.
2025-09-29
ArXiv (prépublication)
arxiv.org
arxiv.org
High-order Component Attribution via Kolmogorov-Arnold Networks
Samy Mammeri
Christian Gagné
Component attribution methods provide insight into how parts of deep learning models, such as convolutional filters and attention heads, inf… (voir plus)luence model predictions. Despite their successes, existing attribution approaches typically assume component effects are additive and independent, neglecting complex interactions among components. Capturing these relations between components is crucial for a better mechanistic understanding of these models. In this work, we improve component attribution (COAR) by replacing the linear counterfactual estimator with a Kolmogorov–Arnold Network (KAN) surrogate fitted to example‑wise perturbation–response data. Then, a symbolic approximation of the learned KAN lets us compute mixed partial derivatives that captures and makes explicit high‑order component interactions that linear methods are missing. These symbolic expressions facilitate future integration with formal verification methods, enabling richer counterfactual analyses of internal model behavior. Preliminary results on standard image classification models demonstrate that our approach improves the accuracy of predicted counterfactuals and enable extraction of higher-order component interactions compared to linear attribution methods.
2025-09-29
NeurIPS.cc/2025/Workshop/MechInterp (poster)
openreview.net
openreview.net
Scaling Synthetic Task Generation for Agents via Exploration
Ram Ramrakhya
Andrew Szot
Omar Attia
Yuhao Yang
Anh Nguyen
Bogdan Mazoure
Zhe Gan
Harsh Agrawal
Alexander T Toshev
Post-Training Multimodal Large Language Models (MLLMs) to build interactive agents holds promise across domains such as computer-use, web na… (voir plus)vigation, and robotics. A key challenge in scaling such post-training is lack of high-quality downstream agentic task datasets with tasks that are diverse, feasible, and verifiable. Existing approaches for task generation rely heavily on human annotation or prompting MLLM with limited downstream environment information, which is either costly or poorly scalable as it yield tasks with limited coverage. To remedy this, we present AutoPlay, a scalable pipeline for task generation that explicitly explores interactive environments to discover possible interactions and current state information to synthesize environment-grounded tasks. AutoPlay operates in two stages: (i) an exploration phase, where an MLLM explorer agent systematically uncovers novel environment states and functionalities, and (ii) a task generation phase, where a task generator leverages exploration trajectories and a set of task guideline prompts as context to synthesize diverse, executable, and verifiable tasks. We show AutoPlay generates 20k tasks across 20 Android applications and 10k tasks across 13 applications Ubuntu applications to train mobile-use and computer-use agents. AutoPlay generated tasks enable large-scale task demonstration synthesis without human annotation by employing an MLLM task executor and verifier. This data enables training MLLM-based UI agents that improve success rates up to
2025-09-29
ArXiv (prépublication)
arxiv.org
arxiv.org
The spatially-resolved effect of mergers on the stellar mass assembly of MaNGA galaxies
E. Angeloudi
Marc Huertas-Company
Jes'us Falc'on-Barroso
Laurence Perreault-Levasseur
Alexandre Adam
A. Boecker
Understanding the origin of stars within a galaxy - whether formed in-situ or accreted from other galaxies (ex-situ) - is key to constrainin… (voir plus)g its evolution. Spatially resolving these components provides crucial insights into a galaxy's mass assembly history. We aim to predict the spatial distribution of ex-situ stellar mass fraction in MaNGA galaxies, and to identify distinct assembly histories based on the radial gradients of these predictions in the central regions. We employ a diffusion model trained on mock MaNGA analogs (MaNGIA), derived from the TNG50 cosmological simulation. The model learns to predict the posterior distribution of resolved ex-situ stellar mass fraction maps, conditioned on stellar mass density, velocity, and velocity dispersion gradient maps. After validating the model on an unseen test set from MaNGIA, we apply it to MaNGA galaxies to infer the spatially-resolved distribution of their ex-situ stellar mass fractions - i.e. the fraction of stellar mass in each spaxel originating from mergers. We identify four broad categories of ex-situ mass distributions: flat gradient, in-situ dominated; flat gradient, ex-situ dominated; positive gradient; and negative gradient. The vast majority of MaNGA galaxies fall in the first category - flat gradients with low ex-situ fractions - confirming that in-situ star formation is the main assembly driver for low- to intermediate-mass galaxies. At high stellar masses, the ex-situ maps are more diverse, highlighting the key role of mergers in building the most massive systems. Ex-situ mass distributions correlate with morphology, star-formation activity, stellar kinematics, and environment, indicating that accretion history is a primary factor shaping massive galaxies. Finally, by tracing their assembly histories in TNG50, we link each class to distinct merger scenarios, ranging from secular evolution to merger-dominated growth.
2025-09-29
ArXiv (prépublication)
arxiv.org
arxiv.org
Visual serial processing deficits explain divergences in human and VLM reasoning
Nicholas Budny
Kia Ghods
Declan Campbell
Raja Marjieh
Amogh Joshi
Sreejan Kumar
Jonathan D. Cohen
Taylor Webb
Thomas L. Griffiths
Why do Vision Language Models (VLMs), despite success on standard benchmarks, often fail to match human performance on surprisingly simple v… (voir plus)isual reasoning tasks? While the underlying computational principles are still debated, we hypothesize that a crucial factor is a deficit in visually-grounded serial processing. To test this hypothesis, we compared human and VLM performance across tasks designed to vary serial processing demands in three distinct domains: geometric reasoning, perceptual enumeration, and mental rotation. Tasks within each domain varied serial processing load by manipulating factors such as geometric concept complexity, perceptual individuation load, and transformation difficulty. Across all domains, our results revealed a consistent pattern: decreased VLM accuracy was strongly correlated with increased human reaction time (used as a proxy for serial processing load). As tasks require more demanding serial processing -- whether composing concepts, enumerating items, or performing mental transformations -- the VLM-human performance gap widens reliably. These findings support our hypothesis, indicating that limitations in serial, visually grounded reasoning represent a fundamental bottleneck that distinguishes current VLMs from humans.
2025-09-29
ArXiv (prépublication)
arxiv.org
arxiv.org
Beyond Embeddings: Interpretable Feature Extraction for Binary Code Similarity
Charles E. Gagnon
Steven H. H. Ding
Philippe Charland
Benjamin Fung
Binary code similarity detection is a core task in reverse engineering. It supports malware analysis and vulnerability discovery by identify… (voir plus)ing semantically similar code in different contexts. Modern methods have progressed from manually engineered features to vector representations. Hand-crafted statistics (e.g., operation ratios) are interpretable, but shallow and fail to generalize. Embedding-based methods overcome this by learning robust cross-setting representations, but these representations are opaque vectors that prevent rapid verification. They also face a scalability-accuracy trade-off, since high-dimensional nearest-neighbor search requires approximations that reduce precision. Current approaches thus force a compromise between interpretability, generalizability, and scalability. We bridge these gaps using a language model-based agent to conduct structured reasoning analysis of assembly code and generate features such as input/output types, side effects, notable constants, and algorithmic intent. Unlike hand-crafted features, they are richer and adaptive. Unlike embeddings, they are human-readable, maintainable, and directly searchable with inverted or relational indexes. Without any matching training, our method respectively achieves 42% and 62% for recall@1 in cross-architecture and cross-optimization tasks, comparable to embedding methods with training (39% and 34%). Combined with embeddings, it significantly outperforms the state-of-the-art, demonstrating that accuracy, scalability, and interpretability can coexist.
2025-09-27
ArXiv (prépublication)
arxiv.org
arxiv.org
CrediBench: Building Web-Scale Network Datasets for Information Integrity
Emma Kondrup
Sebastian Sabry
Hussein Abdallah
Zachary Yang
James Zhou
Kellin Pelrine
Jean-François Godbout
Michael M. Bronstein
Reihaneh Rabbany
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.
2025-09-27
ArXiv (prépublication)
arxiv.org
arxiv.org
CrystalGym: A New Benchmark for Materials Discovery Using Reinforcement Learning
Prashant Govindarajan
Mathieu Reymond
Antoine Clavaud
Mariano Phielipp
Santiago Miret
Sarath Chandar
In silico design and optimization of new materials primarily relies on high-accuracy atomic simulators that perform density functional theor… (voir plus)y (DFT) calculations. While recent works showcase the strong potential of machine learning to accelerate the material design process, they mostly consist of generative approaches that do not use direct DFT signals as feedback to improve training and generation mainly due to DFT's high computational cost. To aid the adoption of direct DFT signals in the materials design loop through online reinforcement learning (RL), we propose CrystalGym, an open-source RL environment for crystalline material discovery. Using CrystalGym, we benchmark common value- and policy-based reinforcement learning algorithms for designing various crystals conditioned on target properties. Concretely, we optimize for challenging properties like the band gap, bulk modulus, and density, which are directly calculated from DFT in the environment. While none of the algorithms we benchmark solve all CrystalGym tasks, our extensive experiments and ablations show different sample efficiencies and ease of convergence to optimality for different algorithms and environment settings. Additionally, we include a case study on the scope of fine-tuning large language models with reinforcement learning for improving DFT-based rewards. Our goal is for CrystalGym to serve as a test bed for reinforcement learning researchers and material scientists to address these real-world design problems with practical applications. We therefore introduce a novel class of challenges for reinforcement learning methods dealing with time-consuming reward signals, paving the way for future interdisciplinary research for machine learning motivated by real-world applications.
2025-09-27
ArXiv (prépublication)
arxiv.org
arxiv.org
Planner Aware Path Learning in Diffusion Language Models Training
Fred Zhangzhi Peng
Zachary Bezemek
Jarrid Rector-Brooks
Shuibai Zhang
Anru R. Zhang
Michael M. Bronstein
Joey Bose
Alexander Tong
Diffusion language models have emerged as a powerful alternative to autoregressive models, enabling fast inference through flexible and para… (voir plus)llel generation paths. This flexibility is enabled by new sampling strategies, or planners, that iteratively choose where to denoise along the sequence rather than sampling uniformly at random. However, by modifying reverse paths, planners introduce a mismatch between the uniformly random denoising paths used during training and the planning-based paths used at inference. In this work, we systematically investigate this mismatch and theoretically show that the standard discrete diffusion training evidence lower bound (ELBO) does not accurately describe a denoiser under non-uniform planning. To bridge this gap, we derive a new Planned Evidence Lower Bound (P-ELBO) that directly incorporates planner-based reverse dynamics into the training objective. Building on this, we propose Planner Aware Path Learning (PAPL), a simple and effective modification of the standard masked discrete diffusion loss that aligns training and inference under planned denoisers. Empirically, PAPL delivers consistent improvements across domains, including a 40% relative gain in protein sequence modeling, up to a 4x improvement in MAUVE for text generation, and a 23% relative gain in HumanEval pass@10 for code generation.
2025-09-27
ArXiv (prépublication)
arxiv.org
arxiv.org
Planning with Unified Multimodal Models
Yihao Sun
Zhilong Zhang
Yang Yu
Pierre-Luc Bacon
With the powerful reasoning capabilities of large language models (LLMs) and vision-language models (VLMs), many recent works have explored … (voir plus)using them for decision-making. However, most of these approaches rely solely on language-based reasoning, which limits their ability to reason and make informed decisions. Recently, a promising new direction has emerged with unified multimodal models (UMMs), which support both multimodal inputs and outputs. We believe such models have greater potential for decision-making by enabling reasoning through generated visual content. To this end, we propose Uni-Plan, a planning framework built on UMMs. Within this framework, a single model simultaneously serves as the policy, dynamics model, and value function. In addition, to avoid hallucinations in dynamics predictions, we present a novel approach self-discriminated filtering, where the generative model serves as a self-discriminator to filter out invalid dynamics predictions. Experiments on long-horizon planning tasks show that Uni-Plan substantially improves success rates compared to VLM-based methods, while also showing strong data scalability, requiring no expert demonstrations and achieving better performance under the same training-data size. This work lays a foundation for future research in reasoning and decision-making with UMMs.
2025-09-27
ArXiv (prépublication)
arxiv.org
arxiv.org