arXiv subjects

Audrey Durand

Publications and source records attributed to Audrey Durand.

At least 19 recordsLinked to original sources

arXiv2026

Training a neural network to rapidly identify candidate gravitational-wave events in the lower mass gap

The physics governing the boundary between the most massive neutron stars (NSs) and the least massive black holes (BHs) is currently uncertain, but could potentially be constrained with new observations. While NSs have been observed with masses up to $\sim2~M_{\odot}$, there is a dearth of electromagnetic observations of compact objects in the $\sim2-5~M_{\odot}$ range, known as the lower mass gap. Recent observations of gravitational-wave (GW) signals from binary mergers detected by the LIGO-Virgo-KAGRA (LVK) collaboration indicate that this gap is likely not empty. Rapidly distinguishing whether a candidate GW event has components in this purported mass gap can indicate the likelihood of a detectable electromagnetic counterpart, and thus inform decisions for follow-up observations. In this work we train a neural network model, GWSkyNet-MassGap, that simultaneously predicts the probability that a candidate merger has a component in the lower mass gap ($P_{\mathrm{MassGap}}$) and the probability that it involves a NS ($P_{\mathrm{NS}}$). We find that the model is able to infer information about the source chirp mass to predict $P_{\mathrm{MassGap}}$ and $P_{\mathrm{NS}}$, leading to correct predictions for high-mass mergers with $\mathcal{M}_c\gtrsim15~M_{\odot}$, but less accurate predictions for lower-mass systems which require knowledge of the binary mass ratio to break the mass degeneracy. For candidate events in the first part of LVK's fourth observing run (O4a), the model has a mean prediction error of 9% for $P_{\mathrm{MassGap}}$ and 6% for $P_{\mathrm{NS}}$. The model could be further developed to rapidly predict the source chirp mass for candidate events in future observing runs.

astro-ph.IM↗

arXiv2026

CanViT: Toward Active-Vision Foundation Models

Active computer vision promises efficient, biologically plausible perception through sequential, localized glimpses, but lacks scalable general-purpose architectures and pretraining pipelines, leaving Active-Vision Foundation Models (AVFMs) underexplored. We introduce CanViT, the first task- and policy-agnostic AVFM. CanViT uses scene-relative RoPE to bind a retinotopic Vision Transformer backbone and a spatiotopic scene-wide latent workspace, the canvas. Efficient interaction with this high-capacity working memory is supported by Canvas Attention, a novel asymmetric cross-attention mechanism. We decouple thinking (backbone-level) and memory (canvas-level), eliminating canvas-side self-attention and fully-connected layers to achieve fast sequential inference and scalability to high output resolutions. We propose a label-free active vision pretraining scheme, policy-agnostic passive-to-active dense latent distillation: reconstructing scene-wide DINOv3 embeddings from sequences of low-resolution glimpses with randomized locations, zoom levels, and lengths. We pretrain CanViT-B from a random initialization on 13.2 million ImageNet-21k scenes--an order of magnitude more than previous active models--and 1 billion random glimpses, in 166 hours on a single H100. On ADE20K segmentation, a frozen CanViT-B achieves 38.5% mIoU in a single low-resolution glimpse, outperforming the best active model's 27.6% with 20x fewer inference FLOPs as well as its FLOP- or input-matched DINOv3 teacher. Given additional glimpses, CanViT-B reaches 45.9% ADE20K mIoU. On ImageNet-1k classification, CanViT-B also sets a new active-vision state of the art, with 84.5% top-1 accuracy after fine-tuning. CanViT generalizes to longer rollouts, larger scenes, and new policies. Our work narrows the wide gap between passive and active computer vision, demonstrating the potential of task- and policy-agnostic AVFM pretraining.

cs.CV↗

arXiv2026

Signal from Structure: Exploiting Submodular Upper Bounds in Generative Flow Networks

Generative Flow Networks (GFlowNets; GFNs) are a class of generative models that learn to sample compositional objects proportionally to their a priori unknown value, their reward. We focus on the case where the reward has a specified, actionable structure, namely that it is submodular. We show submodularity can be harnessed to retrieve upper bounds on the reward of compositional objects that have not yet been observed. We provide in-depth analyses of the probability of such bounds occurring, as well as how many unobserved compositional objects can be covered by a bound. Following the Optimism in the Face of Uncertainty principle, we then introduce SUBo-GFN, which uses the submodular upper bounds to train a GFN. We show that SUBo-GFN generates orders of magnitude more training data than classical GFNs for the same number of queries to the reward function. We demonstrate the effectiveness of SUBo-GFN in terms of distribution matching and high-quality candidate generation on synthetic and real-world submodular tasks.

cs.LG↗

arXiv2025

LLM-as-a-Judge: Toward World Models for Slate Recommendation Systems

Modeling user preferences across domains remains a key challenge in slate recommendation (i.e. recommending an ordered sequence of items) research. We investigate how Large Language Models (LLM) can effectively act as world models of user preferences through pairwise reasoning over slates. We conduct an empirical study involving several LLMs on three tasks spanning different datasets. Our results reveal relationships between task performance and properties of the preference function captured by LLMs, hinting towards areas for improvement and highlighting the potential of LLMs as world models in recommender systems.

cs.IR↗

arXiv2025

Robust Fine-Tuning from Non-Robust Pretrained Models: Mitigating Suboptimal Transfer With Epsilon-Scheduling

Fine-tuning pretrained models is a standard and effective workflow in modern machine learning. However, robust fine-tuning (RFT), which aims to simultaneously achieve adaptation to a downstream task and robustness to adversarial examples, remains challenging. Despite the abundance of non-robust pretrained models in open-source repositories, their potential for RFT is less understood. We address this knowledge gap by systematically examining RFT from such non-robust models. Our experiments reveal that fine-tuning non-robust models with a robust objective, even under small perturbations, can lead to poor performance, a phenomenon that we dub suboptimal transfer. In challenging scenarios (eg, difficult tasks, high perturbation), the resulting performance can be so low that it may be considered a transfer failure. We find that fine-tuning using a robust objective impedes task adaptation at the beginning of training and eventually prevents optimal transfer. However, we propose a novel heuristic, Epsilon-Scheduling, a schedule over perturbation strength used during training that promotes optimal transfer. Additionally, we introduce expected robustness, a metric that captures performance across a range of perturbations, providing a more comprehensive evaluation of the accuracy-robustness trade-off for diverse models at test time. Extensive experiments on a wide range of configurations (six pretrained models and five datasets) show that Epsilon-Scheduling successfully prevents suboptimal transfer and consistently improves expected robustness.

cs.LG↗

arXiv2025

Nested-ReFT: Efficient Reinforcement Learning for Large Language Model Fine-Tuning via Off-Policy Rollouts

Advanced reasoning in LLMs on challenging domains like mathematical reasoning can be tackled using verifiable rewards based reinforced fine-tuning (ReFT). In standard ReFT frameworks, a behavior model generates multiple completions with answers per problem, for the answer to be then scored by a reward function. While such RL post-training methods demonstrate significant performance improvements across challenging reasoning domains, the computational cost of generating completions during training with multiple inference steps makes the training cost non-trivial. To address this, we draw inspiration from off-policy RL, and speculative decoding to introduce a novel ReFT framework, dubbed Nested-ReFT, where a subset of layers of the target model acts as the behavior model to generate off-policy completions during training. The behavior model configured with dynamic layer skipping per batch during training decreases the inference cost compared to the standard ReFT frameworks. Our theoretical analysis shows that Nested-ReFT yields unbiased gradient estimates with controlled variance. Our empirical analysis demonstrates improved computational efficiency measured as tokens/sec across multiple math reasoning benchmarks and model sizes. Additionally, we explore three variants of bias mitigation to minimize the off-policyness in the gradient updates that allows for maintaining performance that matches the baseline ReFT performance.

cs.LG↗

arXiv2025

A Guide to Robust Generalization: The Impact of Architecture, Pre-training, and Optimization Strategy

Deep learning models operating in the image domain are vulnerable to small input perturbations. For years, robustness to such perturbations was pursued by training models from scratch (i.e., with random initializations) using specialized loss objectives. Recently, robust fine-tuning has emerged as a more efficient alternative: instead of training from scratch, pretrained models are adapted to maximize predictive performance and robustness. To conduct robust fine-tuning, practitioners design an optimization strategy that includes the model update protocol (e.g., full or partial) and the specialized loss objective. Additional design choices include the architecture type and size, and the pretrained representation. These design choices affect robust generalization, which is the model's ability to maintain performance when exposed to new and unseen perturbations at test time. Understanding how these design choices influence generalization remains an open question with significant practical implications. In response, we present an empirical study spanning 6 datasets, 40 pretrained architectures, 2 specialized losses, and 3 adaptation protocols, yielding 1,440 training configurations and 7,200 robustness measurements across five perturbation types. To our knowledge, this is the most diverse and comprehensive benchmark of robust fine-tuning to date. While attention-based architectures and robust pretrained representations are increasingly popular, we find that convolutional neural networks pretrained in a supervised manner on large datasets often perform best. Our analysis both confirms and challenges prior design assumptions, highlighting promising research directions and offering practical guidance.

cs.LG↗

arXiv2025

On the Fundamental Limitations of Dual Static CVaR Decompositions in Markov Decision Processes

It was recently shown that dynamic programming (DP) methods for finding static CVaR-optimal policies in Markov Decision Processes (MDPs) can fail when based on the dual formulation, yet the root cause of this failure remains unclear. We expand on these findings by shifting focus from policy optimization to the seemingly simpler task of policy evaluation. We show that evaluating the static CVaR of a given policy can be framed as two distinct minimization problems. We introduce a set of ``risk-assignment consistency constraints'' that must be satisfied for their solutions to match and we demonstrate that an empty intersection of these constraints is the source of previously observed evaluation errors. Quantifying the evaluation error as the \emph{CVaR evaluation gap}, we demonstrate that the issues observed when optimizing over the dual-based CVaR DP are explained by the returned policy having a non-zero CVaR evaluation gap. Finally, we leverage our proposed risk-assignment constraints perspective to prove that the search for a single, uniformly optimal policy on the dual CVaR decomposition is fundamentally limited, identifying an MDP where no single policy can be optimal across all initial risk levels.

cs.LG↗

arXiv2025

Preference-based learning for news headline recommendation

This study explores strategies for optimizing news headline recommendations through preference-based learning. Using real-world data of user interactions with French-language online news posts, we learn a headline recommender agent under a contextual bandit setting. This allows us to explore the impact of translation on engagement predictions, as well as the benefits of different interactive strategies on user engagement during data collection. Our results show that explicit exploration may not be required in the presence of noisy contexts, opening the door to simpler but efficient strategies in practice.

cs.IR↗

arXiv2025

GWSkyNet-Multi II: an updated machine learning model for rapid classification of gravitational-wave events

Multi-messenger observations of gravitational waves and electromagnetic emission from compact object mergers offer unique insights into the structure of neutron stars, the formation of heavy elements, and the expansion rate of the Universe. With the LIGO-Virgo-KAGRA (LVK) gravitational-wave detectors currently in their fourth observing run (O4), it is an exciting time for detecting these mergers. However, assessing whether to follow up a candidate gravitational-wave event given limited telescope time and resources is challenging; the candidate can be a false alert due to detector glitches, or may not have any detectable electromagnetic counterpart even if it is real. GWSkyNet-Multi is a machine learning model developed to facilitate follow-up decisions by providing real-time classification of candidate events, using localization information released in LVK rapid public alerts. Here we introduce GWSkyNet-Multi II, an updated model targeted towards providing more robust and informative predictions during O4 and beyond. Specifically, the model now provides normalized probability scores and associated uncertainties for each of the four corresponding source categories released by the LVK: glitch, binary black hole, neutron star-black hole, and binary neutron star. Informed by explainability studies of the original model, the updated model architecture is also significantly simplified, including replacing input images with intuitive summary values that are more interpretable. For significant event alerts issued during O4a and O4b, GWSkyNet-Multi II produces a prediction that is consistent with the updated LVK classification for 93% of events. The updated model can be used by the community to help make time-critical follow-up decisions.

astro-ph.IM↗

arXiv2025

Adaptive Experiments Under Data Sparse Settings: Applications for Educational Platforms

Adaptive experimentation is increasingly used in educational platforms to personalize learning through dynamic content and feedback. However, standard adaptive strategies such as Thompson Sampling often underperform in real-world educational settings where content variations are numerous and student participation is limited, resulting in sparse data. In particular, Thompson Sampling can lead to imbalanced content allocation and delayed convergence on which aspects of content are most effective for student learning. To address these challenges, we introduce Weighted Allocation Probability Adjusted Thompson Sampling (WAPTS), an algorithm that refines the sampling strategy to improve content-related decision-making in data-sparse environments. WAPTS is guided by the principle of lenient regret, allowing near-optimal allocations to accelerate learning while still exploring promising content. We evaluate WAPTS in a learnersourcing scenario where students rate peer-generated learning materials, and demonstrate that it enables earlier and more reliable identification of promising treatments.

cs.LG↗

arXiv2024

On shallow planning under partial observability

Formulating a real-world problem under the Reinforcement Learning framework involves non-trivial design choices, such as selecting a discount factor for the learning objective (discounted cumulative rewards), which articulates the planning horizon of the agent. This work investigates the impact of the discount factor on the bias-variance trade-off given structural parameters of the underlying Markov Decision Process. Our results support the idea that a shorter planning horizon might be beneficial, especially under partial observability.

cs.AI↗

arXiv2024

Neural Active Learning Meets the Partial Monitoring Framework

We focus on the online-based active learning (OAL) setting where an agent operates over a stream of observations and trades-off between the costly acquisition of information (labelled observations) and the cost of prediction errors. We propose a novel foundation for OAL tasks based on partial monitoring, a theoretical framework specialized in online learning from partially informative actions. We show that previously studied binary and multi-class OAL tasks are instances of partial monitoring. We expand the real-world potential of OAL by introducing a new class of cost-sensitive OAL tasks. We propose NeuralCBP, the first PM strategy that accounts for predictive uncertainty with deep neural networks. Our extensive empirical evaluation on open source datasets shows that NeuralCBP has favorable performance against state-of-the-art baselines on multiple binary, multi-class and cost-sensitive OAL tasks.

cs.LG↗

arXiv2024

Randomized Confidence Bounds for Stochastic Partial Monitoring

The partial monitoring (PM) framework provides a theoretical formulation of sequential learning problems with incomplete feedback. On each round, a learning agent plays an action while the environment simultaneously chooses an outcome. The agent then observes a feedback signal that is only partially informative about the (unobserved) outcome. The agent leverages the received feedback signals to select actions that minimize the (unobserved) cumulative loss. In contextual PM, the outcomes depend on some side information that is observable by the agent before selecting the action on each round. In this paper, we consider the contextual and non-contextual PM settings with stochastic outcomes. We introduce a new class of PM strategies based on the randomization of deterministic confidence bounds. We also extend regret guarantees to settings where existing stochastic strategies are not applicable. Our experiments show that the proposed RandCBP and RandCBPsidestar strategies have favorable performance against state-of-the-art baselines in multiple PM games. To advocate for the adoption of the PM framework, we design a use case on the real-world problem of monitoring the error rate of any deployed classification system.

cs.LG↗

arXiv2023

Association Rules Mining with Auto-Encoders

Association rule mining is one of the most studied research fields of data mining, with applications ranging from grocery basket problems to explainable classification systems. Classical association rule mining algorithms have several limitations, especially with regards to their high execution times and number of rules produced. Over the past decade, neural network solutions have been used to solve various optimization problems, such as classification, regression or clustering. However there are still no efficient way association rules using neural networks. In this paper, we present an auto-encoder solution to mine association rule called ARM-AE. We compare our algorithm to FP-Growth and NSGAII on three categorical datasets, and show that our algorithm discovers high support and confidence rule set and has a better execution time than classical methods while preserving the quality of the rule set produced.

cs.LG↗

arXiv2023

Neural Bandits for Data Mining: Searching for Dangerous Polypharmacy

Polypharmacy, most often defined as the simultaneous consumption of five or more drugs at once, is a prevalent phenomenon in the older population. Some of these polypharmacies, deemed inappropriate, may be associated with adverse health outcomes such as death or hospitalization. Considering the combinatorial nature of the problem as well as the size of claims database and the cost to compute an exact association measure for a given drug combination, it is impossible to investigate every possible combination of drugs. Therefore, we propose to optimize the search for potentially inappropriate polypharmacies (PIPs). To this end, we propose the OptimNeuralTS strategy, based on Neural Thompson Sampling and differential evolution, to efficiently mine claims datasets and build a predictive model of the association between drug combinations and health outcomes. We benchmark our method using two datasets generated by an internally developed simulator of polypharmacy data containing 500 drugs and 100 000 distinct combinations. Empirically, our method can detect up to 72% of PIPs while maintaining an average precision score of 99% using 30 000 time steps.

cs.LG↗

arXiv2023

Interpret Your Care: Predicting the Evolution of Symptoms for Cancer Patients

Cancer treatment is an arduous process for patients and causes many side-effects during and post-treatment. The treatment can affect almost all body systems and result in pain, fatigue, sleep disturbances, cognitive impairments, etc. These conditions are often under-diagnosed or under-treated. In this paper, we use patient data to predict the evolution of their symptoms such that treatment-related impairments can be prevented or effects meaningfully ameliorated. The focus of this study is on predicting the pain and tiredness level of a patient post their diagnosis. We implement an interpretable decision tree based model called LightGBM on real-world patient data consisting of 20163 patients. There exists a class imbalance problem in the dataset which we resolve using the oversampling technique of SMOTE. Our empirical results show that the value of the previous level of a symptom is a key indicator for prediction and the weighted average deviation in prediction of pain level is 3.52 and of tiredness level is 2.27.

cs.LG↗

arXiv2022

Cambrian Explosion Algorithm for Multi-Objective Association Rules Mining

Association rule mining is one of the most studied research fields of data mining, with applications ranging from grocery basket problems to highly explainable classification systems. Classical association rule mining algorithms have several flaws especially with regards to their execution times, memory usage and number of rules produced. An alternative is the use of meta-heuristics, which have been used on several optimisation problems. This paper has two objectives. First, we provide a comparison of the performances of state-of-the-art meta-heuristics on the association rule mining problem. We use the multi-objective versions of those algorithms using support, confidence and cosine. Second, we propose a new algorithm designed to mine rules efficiently from massive datasets by exploring a large variety of solutions, akin to the explosion of species diversity of the Cambrian Explosion. We compare our algorithm to 20 benchmark algorithms on 22 real-world data-sets, and show that our algorithm present good results and outperform several state-of-the-art algorithms.

cs.NE↗