Publications

Recent progress of large-scale biomarker consortia and paths forward in biomarker development for autism
Jason W. Griffin
Brianna Cairney
William E. Carson, IV
Lacey Chetcuti
Adrien E.E. Dubois
Guillaume Dumas
Sébastien Jacquemont
Shafali Jeste
Jacob P. Momsen
Adam J. Naples
James C. McPartland
2026-07-31
Biological Psychiatry (publié)
doi.org
Hierarchical Clustering of Networks via Hierarchical Distance Matrices
Li Chen
Nathaniel Josephs
Eric D. Kolaczyk
Lizhen Lin
Clustering populations of networks while recovering their latent hierarchical organization is a fundamental yet largely unexplored problem i… (voir plus)n network analysis. To formalize this, we introduce the Hierarchical Distance Matrix, a specific class of population-level distance matrices that encodes latent hierarchical organization through recursively nested distance separation, accommodating unbalanced tree depths. Building on this framework, we propose a fully data-driven top-down procedure: network hierarchical clustering based on two-sample testing (NHC-TST). The algorithm recursively splits networks via spectral clustering and uses a graph-based two-sample stopping rule. The procedure adaptively determines the branching structure without requiring prior knowledge of the number of clusters or tree depth. Theoretically, we establish exact recovery of the population-level hierarchical structure and statistical consistency in the empirical procedure. Simulation studies demonstrate highly accurate recovery of both cluster memberships and hierarchical relationships across a wide range of settings. Applied to a global migration dataset, NHC-TST uncovers interpretable multi-resolution temporal structures that are not revealed by conventional flat clustering approaches.
2026-07-30
arXiv (prépublication)
doi.org
arxiv.org
One Flow-Transformer for Imagination and Control
Rabiul Awal
Jinseong Jeong
Ankur Sikarwar
Parisa Kordjamshidi
Andrii Zadaianchuk
Sai Rajeswar
Paul Hongsuck Seo
Aishwarya Agrawal
Diffusion and flow models are effective world models for visual reinforcement learning, but existing agents treat them as black-box simulato… (voir plus)rs, leaving the backbone’s representations unused for control. We introduce DRIFT, an online agent in which a single Flow-Transformer serves as both world model and policy backbone, trained from scratch. We find that denoising features alone are suboptimal for control; DRIFT bridges this gap with a next-latent prediction objective that gives the backbone an explicit dynamics signal. Shortcut flow matching reduces imagination to a single denoising step per frame. Across Atari 100k, Craftium, and Crafter, DRIFT is competitive with both latent-dynamics and diffusion world-model baselines.
2026-07-30
WMW @ Reinforcement Learning Conference (publié)
openreview.net
openreview.net
TOOD: Task-Aware Out-of-Distribution Score Calibration for Continual Learners
Mostafa Elaraby
Samer B. Nashed
Liam Paull
The primary challenge of continual learning (CL) systems is to learn new tasks while remaining performant on previously learned tasks. A sim… (voir plus)ilarly important though less well-studied aspect of CL systems is their ability to distinguish inputs that are unlikely to come from within the set of tasks the system has already encountered, often called out-of-distribution (OOD) detection. This paper presents several findings related to the dynamics of OOD detection in CL systems, causes of performance degradation over time which we call OOD forgetting (OODF), and proposed mitigation strategies for this degradation. Chiefly, we find the unintuitive result that OODF is only weakly anti-correlated with classification performance on previous tasks, suggesting that the underlying mechanisms producing OODF are distinct. Moreover, this effect is observed for both energy-based and feature-based OOD detection methods. Energy-based detectors suffer a drop in logit scale as additional tasks are learned, which we term the Confidence Gap, while feature-based detectors also degrade under a complementary effect we call Manifold Crowding. Motivated by these observations, we propose TOOD, a training-free post-hoc method that decomposes logits into per-task energy scores and re-calibrates them using replay-buffer statistics. Experiments on CIFAR-10, CIFAR-100, and a 100-task ImageNet-1K stream show that TOOD improves OOD detection performance over uncalibrated energy in most settings and ranks first or second in nine of ten CIFAR configurations, with the largest gains when the confidence gap is most severe. These results suggest that a substantial portion of OOD deterioration in continual learning arises from score miscalibration rather than from a complete loss of discriminative structure.
2026-07-30
arXiv (prépublication)
doi.org
arxiv.org
Combining sampling and attractor dynamics in spiking models of head direction systems
Vojko Pjanovic
Jacob A. Zavatone-Veth
Paul Masset
Sander W. Keemink
Michele Nardin
Neural populations can maintain stable representations of navigation-related variables while integrating uncertain sensory signals. Experime… (voir plus)ntal evidence showed that the precision of head-direction (HD) representations in flies and mice depends on the reliability of sensory cues, highlighting the influence of input uncertainty in attractor-based neural circuits. How do neural dynamics maintain stability while computing under uncertainty? Here, we propose a spiking neural network that unifies two principles - stability through attraction and uncertainty through fluctuation - and reinterpret the HD circuit as an uncertainty-aware integrator rather than a deterministic compass. Specifically, the network uses sampling-based probabilistic inference, where a neural population represents input uncertainty by rapidly fluctuating among likely hypotheses about the world while preserving a stable representation of head direction along an attractor manifold. This formulation suggests why a classical HD "bump" becomes less precise, namely due to rapid fluctuations, reflecting the uncertainty in angular velocity inputs. Our implementation yields experimentally testable predictions: correlated subthreshold voltage fluctuations, multi-timescale nonlinear interaction patterns, and characteristic statistics of bump movement. By combining probabilistic inference with attractor dynamics within one single circuit, our framework suggests how neural populations across species can represent an estimate and its uncertainty through fluctuations while maintaining stability, which could be a general principle for uncertainty-aware computation in noisy biological systems.
2026-07-29
PLOS Computational Biology (publié)
doi.org
Deep learning segmentation with curvature consistency reveals astrocyte nanostructure across species
Albert Hiu Ka Fok
Yanan Wang
Megan Ng
Tabish Syed
Timothy L. H. Wong
Chris K. Salmon
Tomos Salathiel
William Grondin-Eddy
Lanxin Fan
Kaleem Siddiqi
Keith K. Murai
Cellular function depends on the precise deployment and distribution of nanoscale structures, but these features remain difficult to measure… (voir plus) and compare between cells and datasets. This challenge is pronounced for astrocytes, whose intricate nanostructures interface with neurons, glia, and vasculature, to control brain development, synaptic development/plasticity, homeostasis, and responses to injury/disease. Here, we developed deep learning approaches with curvature consistency for automated astrocyte segmentation across volume electron microscopy datasets, reducing reconstruction time from manual or semi-automatic methods by 12-fold and enabling brain region and cross-species interrogation of astrocytic nanoarchitecture. This allowed us to uncover organizing principles and motifs alongside ultrastructural divergence between species. While both species exhibit a wide but shallow topological network, marmoset astrocytes display increased process thickness and branching. We further identified extrasynaptic neuronal engulfment and a robust astrocytic endosomal system across species and brain regions. Together, our findings demonstrate previously inaccessible structural principles of astrocytes, offering a framework for understanding structure-function relationships in the central nervous system.
2026-07-29
Cell Reports (publié)
doi.org
DeltaServe: Host-Agnostic Co-Serving of Inference and Fine-Tuning for LLMs
Jiaxuan Chen
Jianshu She
Ye Yuan
Rajat Ghosh
Karan Gupta
Qirong Ho
Xue Liu
Oana Balmau
LLM serving systems are provisioned for peak load to meet strict latency targets, leaving substantial GPU compute idle whenever traffic fall… (voir plus)s below peak. We present DeltaServe, a host-agnostic co-serving design that converts this idle inference capacity into LoRA fine-tuning throughput while preserving inference service-level objectives (SLOs). DeltaServe integrates with existing inference engines through a compact hook interface that requires only multi-LoRA batching support. It exploits the shared execution structure of inference prefill and LoRA fine-tuning forward passes, and uses an SLO-aware scheduler to admit and execute fine-tuning only when sufficient inference headroom is available. The scheduler is driven by a CUDA-graph-aware latency model calibrated offline and refined online. We integrate DeltaServe with vLLM, SGLang, and S-LoRA. On a production trace from Company X, DeltaServe on vLLM delivers 2.9x higher fine-tuning throughput than LLMStation at 100% inference SLO compliance, versus 85% for LLMStation. It also achieves 39% higher fine-tuning throughput than a baseline running vLLM+torchtune, using no additional hardware and maintaining full SLO compliance.
2026-07-29
arXiv (prépublication)
doi.org
arxiv.org
Finding Change in Satellite Archives from Text: How to Combine Before-and-After Images Efficiently
Simon Roy
Mark Bong
Giovanni Beltrame
Operational Earth observation increasingly calls for answering queries such as ``find the image pairs where a new building appeared.''This m… (voir plus)eans searching an archive of before-and-after (bi-temporal) satellite image pairs and ranking each pair by how well it matches a natural-language description of the change. The component that performs this match, the fusion module that combines the ``before''and ``after''views, must be run at query time across many candidate pairs, so its speed largely sets the cost of every search. We present a controlled comparison of how to build that module. Using one fixed image encoder (a frozen CLIP model) and one training recipe for all variants, we evaluate eight designs drawn from three families: attention, state-space models (Mamba), and learned compression (our Temporal Bottleneck Fusion, TBF). Each design is tested on two benchmarks (LEVIR-CC and Dubai-CC) with ten random seeds, so the reported differences are statistically grounded. We outline three findings: first, a training-free two-stage search (a cheap difference model that shortlists candidates, followed by attention fusion that re-ranks them) matches or exceeds full-fusion recall on LEVIR-CC while cutting query cost
2026-07-29
arXiv (prépublication)
doi.org
arxiv.org
pLM representations unlock metagenomic space beyond homology
Lola Le Breton
David Heurtel-Depeiges
Douglas C. Millar
Lara E. Zetzsche
Robert M. Vernon
Christopher J. Langmead
Sarath Chandar
Quentin Fournier
Metagenomic sequencing has uncovered billions of proteins from uncultured microorganisms, vastly expanding the known protein space. Yet most… (voir plus) remain functionally inaccessible because existing annotation methods depend on close homologs or accurate structure predictions. Here, we show that protein language models (pLMs) can unlock this diversity only when their training data are appropriately curated. We introduce Residue Embedding Diversity (RED), a metric for protein quality assessment orders of magnitude cheaper than likelihood, and a calibration task that measures model alignment with natural evolutionary distributions. We discover a fundamental trade-off between evolutionary calibration and structural modeling, establishing training data composition as a primary determinant of pLM behavior. Finally, we successfully retrieve diverse enzyme candidates from billions of metagenomic sequences and validate their expression in vivo .
2026-07-29
bioRxiv (prépublication)
doi.org
Transparent Envelope-Based Automatic Detection of Obstructive Sleep Apnea
Eva Almeida
Margarida Carvalho
João Pedrosa
Gonzalo C. Gutiérrez-Tobal
Daniela Ferreira-Santos
This study presents an interpretable signal processing approach for the automatic detection of apneas and hypopneas from nasal pressure (NP)… (voir plus) signals. Unlike black-box models, the proposed algorithm relies on physiologically grounded mathematical operations and adaptive amplitude thresholds, ensuring every detection is clinically traceable. The method achieved a sensitivity of 63.8%, a mean Jaccard index of 60.6%, and a temporal overlap of 70.5%. The algorithm outperformed a Deep Learning benchmark (Mr-ResNet) in sensitivity by approximately 20% under subject-independent validation, demonstrating superior resilience to clinical signal noise. Integrating sleep stage information improved precision to 50.5% and enhanced Apnea-Hypopnea Index (AHI) severity classification. These findings demonstrate that interpretable methods can provide more stable and auditable outcomes than complex neural networks, offering a computationally efficient alternative forautomated sleep analysis and integration into wearable diagnostic devices.
2026-07-29
IEEE Conference on Body Sensor Networks (publié)
openreview.net
openreview.net
Beyond Naïve Prompting: Strategies for Improved Context-aided Forecasting with LLMs
Arjun Ashok
Andrew Robert Williams
Vincent Zhihao Zheng
Irina Rish
Nicolas Chapados
Étienne Marcotte
Valentina Zantedeschi
Alexandre Drouin
Real-world forecasting requires models to integrate not only historical data but also relevant contextual information provided in textual fo… (voir plus)rm. While large language models (LLMs) show promise for context-aided forecasting, critical challenges remain: we lack diagnostic tools to understand failure modes, performance remains far below their potential, and high computational costs limit practical deployment. We introduce a unified framework of four strategies that address these limitations along three orthogonal dimensions: model diagnostics, accuracy, and efficiency. Through extensive evaluation across model families from small open-source models to frontier models including Gemini, GPT, and Claude, we uncover both fundamental insights and practical solutions. Our findings span three key dimensions: diagnostic strategies reveal the “Execution Gap” where models correctly explain how context affects forecasts but fail to apply this reasoning; accuracy-focused strategies achieve substantial performance improvements of 25-50%; and efficiency-oriented approaches show that adaptive routing between small and large models can approach large model accuracy on average while significantly reducing inference costs. These orthogonal strategies can be flexibly integrated based on deployment constraints, providing practitioners with a comprehensive toolkit for practical LLM-based context-aided forecasting. Code is made available at https://github.com/ashok-arjun/beyond-naive-prompting.
2026-07-27
Transactions on Machine Learning Research (accepté)
doi.org
openreview.net
Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systems
Marylou Fauchard
Florian Carichon
Margarida Carvalho
Golnoosh Farnadi
Large Language Models (LLMs)-powered multi-agent systems are increasingly deployed in mixed-motive environments, where agents operate under … (voir plus)asymmetric information and strategic deception due to conflicting or hidden objectives. In these settings, misalignment with collective goals becomes a central concern. We propose a novel framework for evaluating objective misalignment using the social deduction game Werewolf, modifying the objective of a single agent while preserving its assigned role. Across LLMs from four different model families and sizes, four player roles, and three objective formulations, we introduce a dual analysis of the agents' internal reasoning and their public cheap-talk behavior (i.e costless, non-binding communication that does not directly affect the agents' utilities), complemented by an analysis of game outcomes. Our results show that objective misalignment undermines outcomes in inherently adversarial environments, an effect exacerbated by asymmetric information and specialized roles. While compromised agents consistently develop distinct objective-dependent reasoning strategies, these adaptations remain largely invisible in their public behavior. More broadly, our findings suggest that even subtle objective misalignment can profoundly affect collective decision-making, highlighting the need for effective mitigation strategies for LLM-based multi-agent systems.
2026-07-27
arXiv (prépublication)
doi.org
arxiv.org