Publications

A Universal Source-Free Class Unlearning Framework via Synthetic Embeddings
Zahra Dehghani Tafti
Pablo Piantanida
Mohammadhadi Shateri
Class unlearning in neural classifiers refers to selectively removing the model’s ability to recognize a target (forget) class by reshapin… (see more)g the decision boundaries. This is essential when taxonomies change, labels are corrected, or legal or ethical requirements mandate class removal. The objective is to preserve performance on the remaining (retain) classes while avoiding costly full retraining. Existing methods generally require access to the source, i.e., forget/retain data or a relevant surrogate dataset. This dependency limits their applicability in scenarios where access to source data is restricted or unavailable. Even the recent source-free class unlearning methods rely on generating samples in the data space, which is computationally expensive and not even essential for doing class unlearning. In this work, we propose a novel source-free class unlearning framework that enables existing unlearning methods to operate using only the deployed model. We show that, under assumptions on the forget loss with respect to logits, class unlearning can be performed source-free for any given neural classifier by utilizing randomly generated samples within the classifier’s intermediate space. Specifically, randomly generated embeddings pseudo-labeled by the model as belonging to the forget or retain classes can support effective source-free unlearning. Our analysis further shows that, under conditions on the forget loss and synthetic forget embeddings, minimizing the forget loss induces expected logit shifts consistent with class unlearning, without requiring a specific parametric form of the embedding distribution. We validate our framework on four backbone architectures, ResNet-18, ResNet-50, ViT-B/16, and Swin-T, across three benchmark datasets, CIFAR-10, CIFAR-100, and TinyImageNet. Our experimental results show that existing class unlearning methods can operate within our source-free framework, with minimal impact on their forgetting efficacy and retain class accuracy. The code is available at https://github.com/Yasaman-dt/Source_Free_Class_Unlearning.
2026-05-18
Transactions on Machine Learning Research (accepted)
openreview.net
openreview.net
Factorized and Vectorized Execution: Optimizing Analytical and Semantic Queries over Relations
Sunny Yasser
Anas Dorbani
Amine Mhedhbi
Many-to-many joins are central to analytical and semantic workloads such as fraud detection, network analysis, and recommendation, where ins… (see more)ights arise from relationships between entities. These workloads often suffer from an explosion of intermediate results, sometimes orders of magnitude larger than the inputs. Factorized representations address this problem by exploiting conditional independence among attributes to encode intermediates more compactly. In some cases, they can reduce the output size asymptotically below the worst-case output size. However, adopting factorization in modern vectorized query processors remains challenging: factorized representations are hierarchical, whereas vectorized execution is built around flat, block-oriented processing. Prior approaches either rely on full materialization or support only restricted factorization layouts, sacrificing much of the benefits of both factorization and vectorization. We present FFX, a novel engine for F ast F actorized e X ecution. FFX is the first pipelined engine to support arbitrary factorization schemes while preserving full vectorization. The engine introduces packed factorized vectors and operators that maintain cache-friendly, contiguous layouts. Beyond analytics, FFX also co-optimizes semantic operators by serializing factorized intermediates into compact prompts for large language models (LLMs), substantially reducing token usage and inference cost while maintaining output quality and, in some cases, improving it. Together, these contributions enable efficient execution of join-heavy analytical queries, including queries augmented with semantic operators.
2026-05-17
Proceedings of the ACM on Management of Data (published)
doi.org
Modelling Customer Trajectories with Reinforcement Learning for Practical Retail Insights
Ken Ming Lee
Paul Barde
Maxime C. Cohen
Derek Nowrouzezahrai
Understanding customer movement within retail spaces is essential for optimizing store layouts. Real-world trajectory data can provide highl… (see more)y accurate insights, but collecting it is costly and often infeasible for many retailers. Heuristics such as Travelling Salesman Problem (TSP) and Probabilistic Nearest Neighbours (PNN) are commonly used as inexpensive approximations, but actual customer trajectories deviate by an average of 28% from shortest paths, highlighting a tradeoff between accuracy and practicality. We propose an agent-based modelling framework that casts customer trajectory prediction as a maximum entropy reinforcement learning (RL) problem, balancing reward maximization with stochasticity to better reflect customers with bounded rationality. Using real-world trajectory data from a convenience store, we show that RL-generated trajectories align more closely with customer behaviour than TSP and PNN, providing more accurate estimates of impulse purchase rates and shelf traffic densities. Furthermore, only RL-based predictions yield repositioning decisions for impulse products that align with those derived from actual trajectory data, resulting in comparable estimated profit gains. Our work demonstrates that RL provides a practical, behaviourally grounded alternative that bridges the gap between oversimplified heuristics and data-intensive approaches, making accurate layout optimization more accessible. To encourage further research, the source code is available on GitHub.
2026-05-17
arXiv (preprint)
doi.org
arxiv.org
PRISM: High-Resolution & Precise Counterfactual Medical Image Generation using Language-guided Stable Diffusion
Amar Kumar
Anita Kriz
Mohammad Havaei
Tal Arbel
Developing reliable and generalizable deep learning systems for medical imaging faces significant obstacles due to spurious correlations, da… (see more)ta imbalances, and limited text annotations in datasets. Addressing these challenges requires architectures robust to the unique complexities posed by medical imaging data. The rapid advancements in vision-language foundation models within the natural image domain prompt the question of how they can be adapted for medical imaging tasks. In this work, we present PRISM, a framework that leverages foundation models to generate high-resolution, language-guided medical image counterfactuals using Stable Diffusion. Our approach demonstrates unprecedented precision in selectively modifying spurious correlations (the medical devices) and disease features, enabling the removal and addition of specific attributes while preserving other image characteristics. Through extensive evaluation, we show how PRISM advances counterfactual generation and enables the development of more robust downstream classifiers for clinically deployable solutions. To facilitate broader adoption and research, we make our code publicly available at https://github.com/Amarkr1/PRISM.
2026-05-17
Medical Imaging with Deep Learning (published)
doi.org
proceedings.mlr.press
Revisiting Age of Acquisition in Curriculum Learning: Disentangling Lexical Features and Semantic Structure
Ian Gifford
Aaron Shah
Catherine Chen
Taimaa Kassab Bachi
Eva Portelance
Previous work has found that ordering training data by children’s Age of Acquisition (AoA) for words increases the stability of distributi… (see more)onal word embeddings, suggesting that early-learned words play a privileged role in shaping semantic structure. In this study, we determine whether AoA itself drives these effects, or whether they emerge from correlated lexical factors such as frequency, concreteness, and phonological complexity. Using incremental Word2Vec training, we construct curricula ordered by AoA and by individual lexical features, while systematically controlling for vocabulary growth and deterministic ordering effects. We show that AoA-ordered curricula produce greater early-phase stability than shuffled baselines, even under controlled exposure conditions. We find that the advantage observed with AoA can be largely explained by correlated factors like overall word frequency. Despite limited gains on general similarity benchmarks, AoA-ordered embeddings outperform shuffled embeddings on a proxy domain-specific task: predicting human AoA norms. This advantage persists after debiasing timestamp effects, implying that AoA curricula induce developmentally meaningful semantic structure.
2026-05-17
CoNLL @ Association for Computational Linguistics (published)
openreview.net
openreview.net
Scalable Environments Drive Generalizable Agents
Jiayi Zhang
Fanqi Kong
Guibin Zhang
Maojia Song
Zhaoyang Yu
Jianhao Ruan
Jinyu Xiang
Bang Liu
Chenglin Wu
Yuyu Luo
Generalizable agents should adapt to diverse tasks and unseen environments beyond their training distribution. This position paper argues th… (see more)at such generalization requires environment scaling: expanding the distribution of executable rule-sets that agents interact with, rather than only increasing trajectories or tasks within fixed benchmarks. Current scaling practices largely focus on collecting more experience or broader task sets under fixed interaction rules, leaving agents brittle when underlying interfaces, dynamics, observations, or feedback signals change. The core challenge is therefore a world-level distribution shift: agents need systematic exposure to environments with meaningfully different executable rule-sets. To clarify this challenge, we propose a unified taxonomy that separates trajectory scaling, task scaling, and environment scaling by their primary deliverables and by what changes in the executable rule-set. Building on this taxonomy, we synthesize construction paradigms for scalable environments, contrasting programmatic generators that prioritize controllability and verifiability with generative world models that offer broader coverage and open-endedness. We further outline how environment scaling can be coupled with stateful learning mechanisms, emphasizing learned update rules for cross-environment adaptation. We conclude by discussing alternative perspectives and argue that scalable environments provide the essential substrate for measurable and controllable progress toward robust general agents.
2026-05-17
arXiv (preprint)
doi.org
arxiv.org
Applying graph neural networks to predict fungal disease occurrences in precision agriculture
Stéphane Samson
Étienne Lord
Odile Carisse
Vladimir Makarenkov
Abstract Purpose Fungal diseases remain among the leading causes of global crop losses, with management still heavily reliant on fungicide a… (see more)pplications. While traditional decision support systems and machine learning models offer valuable predictive insights, they often overlook the spatial and relational dynamics underlying pathogen spread. This study evaluates the feasibility and advantages of Graph Neural Networks (GNNs) for predicting fungal disease occurrence in three key crops—onion ( Botrytis squamosa ), lettuce ( Botrytis lactucae ), and carrot ( Cercospora carotae )—to enhance precision agriculture decision-making. Methods Field observations from farms in southern Quebec were used to build plant-level graphs, with nodes representing plants enriched by biological and weather features, and edges defined by spatial proximity. Graph convolutional networks were trained for binary fungal disease occurrence classification and benchmarked against machine learning and deep learning baselines. Graph augmentation techniques and robustness tests under missing and noisy features were applied to assess GNN’s stability. Results Across the three pathosystems, GNNs achieved the strongest overall predictive performance. For onions ( B. squamosa ), Random Forest slightly outperformed the GNN on the complete feature set (accuracy = 76.4% and F1-score = 0.77); here, the GNN provided lower but comparable metric scores (accuracy = 74.8% and F1-score = 0.73). For lettuce ( B. lactucae ), the GNN achieved the highest metric scores with the accuracy of 90.4% and F1-score of 0.90, surpassing all other baselines. For carrot ( C. carotae ), GNNs reached the accuracy of 75.8% and F1-score of 0.77, clearly outperforming Decision Tree, Random Forest, k-NN, and Feed-Forward Neural Networks (FFNs). Graph augmentation further improved the GNN results: random walk sampling increased the model’s accuracy on onion data to 79.3% and F1-score to 0.79, and on lettuce data to 93.9% and to 0.94, respectively, while node/edge perturbation improved the model’s accuracy on carrot data to 78.6% and F1-score to 0.80. Furthermore, the results of the robustness experiments suggest that GNNs can still track overall field-level infection trends with up to 75% of features masked or 50% replaced by noise. Conclusion GNNs offer clear advantages for fungal disease occurrence prediction by incorporating spatial and relational plant patterns, thus improving both the accuracy and robustness of predicted outcomes.
2026-05-15
Precision Agriculture (published)
doi.org
High-dimensional Limit of SGD for Diagonal Linear Networks
Begoña García Malaxechebarría
Courtney Paquette
Maryam Fazel
Dmitriy Drusvyatskiy
Understanding the behavior of stochastic gradient methods is a central problem in modern machine learning. Recent work has highlighted diago… (see more)nal linear networks as a simplified yet expressive setting for analyzing the optimization and generalization properties of neural models. In this work, we show that in the high-dimensional regime, stochastic gradient descent on diagonal linear networks is well-approximated by continuous dynamics governed by a stochastic differential equation (SDE), which explicitly decouples the drift from the gradient noise. We further derive a deterministic partial differential equation whose solution propagates the relevant state of the iterates and characterizes the time evolution of a broad class of observable statistics, including the risk, curvature, and other metrics for optimality. Finally, we show that, under a suitable parametrization, the stochastic dynamics are globally well posed and converge exponentially fast to zero risk with high probability, yielding a fully explicit non-asymptotic description of their long-time behavior. Numerical simulations corroborate our theoretical findings.
2026-05-15
arXiv (preprint)
doi.org
arxiv.org
A stratified approach for heterogeneous data fusion using polygon generation, deep learning and ensemble modeling
Mohamed Elhefnawy
Ahmed Ragab
Nicolas Pelletier
Jean-Martin Lussier
Mouloud Amazouz
The widespread adoption of digitalization across various industries has resulted in the accumulation of vast amounts of data from diverse so… (see more)urces, offering opportunities to analyze complex phenomena. However, such analysis remains challenging due to the heterogeneity in data types, structures, formats, sampling frequencies and other factors. While numerous artificial intelligence (AI) techniques exist to analyze these heterogeneous datasets, no single AI technique can effectively handle all types of heterogeneous data acquired from different sources, while accurately predicting multiple outputs. To address this challenge, this paper proposes an ensemble learning approach based on diversified regression techniques to predict multiple continuous outputs. This approach captures the data distribution of the targeted phenomena from different perspectives. A novel stratified modeling technique is introduced, in which the data are first classified by predicting stratification labels before undergoing the regression modeling. This stratification is achieved using an innovative fusion approach based on a polygon generation representation technique, effectively breaking down the primary complex discriminative problem into smaller and more manageable subproblems. The effectiveness of the proposed method is validated using a dataset collected from harvester machinery in the forest industry. The method predicts key outputs such as merchantable wood volume, the log count, and the proportion of hardwood and softwood of different sizes for predetermined forest plots. The results demonstrate that the performance of this stratified approach outperforms comparable methods from the literature, demonstrating its superior performance.
2026-05-15
Ecological Informatics (published)
doi.org
A stratified approach for heterogeneous data fusion using polygon generation, deep learning and ensemble modeling
Mohamed Elhefnawy
Ahmed Ragab
Nicolas Pelletier
Jean-Martin Lussier
Mouloud Amazouz
The widespread adoption of digitalization across various industries has resulted in the accumulation of vast amounts of data from diverse so… (see more)urces, offering opportunities to analyze complex phenomena. However, such analysis remains challenging due to the heterogeneity in data types, structures, formats, sampling frequencies and other factors. While numerous artificial intelligence (AI) techniques exist to analyze these heterogeneous datasets, no single AI technique can effectively handle all types of heterogeneous data acquired from different sources, while accurately predicting multiple outputs. To address this challenge, this paper proposes an ensemble learning approach based on diversified regression techniques to predict multiple continuous outputs. This approach captures the data distribution of the targeted phenomena from different perspectives. A novel stratified modeling technique is introduced, in which the data are first classified by predicting stratification labels before undergoing the regression modeling. This stratification is achieved using an innovative fusion approach based on a polygon generation representation technique, effectively breaking down the primary complex discriminative problem into smaller and more manageable subproblems. The effectiveness of the proposed method is validated using a dataset collected from harvester machinery in the forest industry. The method predicts key outputs such as merchantable wood volume, the log count, and the proportion of hardwood and softwood of different sizes for predetermined forest plots. The results demonstrate that the performance of this stratified approach outperforms comparable methods from the literature, demonstrating its superior performance.
2026-05-15
Ecological Informatics (published)
doi.org
TabPFN-MT: A Natively Multitask In-Context Learner for Tabular Data
Cormac Cureton
Narges Armanfard
Prior-Data Fitted networks (PFNs) have been very successful in tabular contexts, handling prediction tasks in context. However, they are des… (see more)igned for single-task inference, meaning that predicting several target values within a context requires repeated forward calls and precludes inter-task information sharing. We propose TabPFN-MT, which is trained on an expanded multi-target synthetic prior to capture inter-task dependencies in context. This model uses an expanded
2026-05-15
arXiv (preprint)
doi.org
arxiv.org
Navigating Potholes with Geometry-Aware Sharpness Minimization
Simon Dufort-Labbé
Mehrab Hamidi
Razvan Pascanu
Ioannis Mitliagkas
Damien Scieur
Aristide Baratin
Sharpness-aware minimization (SAM) encourages flat minima by perturbing parameters along directions of high loss curvature, but treats all p… (see more)arameter directions uniformly, ignoring the underlying loss geometry. We introduce LLQR+SAM, which combines SAM with a learned preconditioner obtained from the recently proposed LLQR framework, a second-order method that recasts steepest descent as a layerwise linear-quadratic regulator problem. The preconditioner is updated sparsely and maintained as a slow exponential moving average, so it captures a smoothed, low-resolution picture of the loss landscape geometry. The SAM perturbation then operates on top of this learned geometry, probing curvature at a faster timescale. We show that this two-timescale structure is not merely a computational convenience: theoretically, the preconditioner amplifies the SAM escape signal in directions that are flat under the average geometry but locally sharp (potholes). Wide, flat basins, by contrast, remain stable. Empirically, LLQR+SAM gives consistent gains over both SAM and LLQR alone across standard vision and sequence modeling benchmarks, supporting the view that slow learned geometry and fast sharpness correction are genuinely complementary.
2026-05-14
arXiv (preprint)
doi.org
arxiv.org