Publications

The Intricate Dance of Prompt Complexity, Quality, Diversity and Consistency in T2I Models
Zhang Xiaofeng
Aaron Courville
Michal Drozdzal
Adriana Romero-Soriano
Text-to-image (T2I) models offer great potential for creating virtually limitless synthetic data, a valuable resource compared to fixed and … (voir plus)finite real datasets. Previous works evaluate the utility of synthetic data from T2I models on three key desiderata: quality, diversity, and consistency. While prompt engineering is the primary means of interacting with T2I models, the systematic impact of prompt complexity on these critical utility axes remains underexplored. In this paper, we first conduct synthetic experiments to motivate the difficulty of generalization w.r.t. prompt complexity and explain the observed difficulty with theoretical derivations. Then, we introduce a new evaluation framework that can compare the utility of real data and synthetic data, and present a comprehensive analysis of how prompt complexity influences the utility of synthetic data generated by commonly used T2I models. We conduct our study across diverse datasets, including CC12M, ImageNet-1k, and DCI, and evaluate different inference-time intervention methods. Our synthetic experiments show that generalizing to more general conditions is harder than the other way round, since the former needs an estimated likelihood that is not learned by diffusion models. Our large-scale empirical experiments reveal that increasing prompt complexity results in lower conditional diversity and prompt consistency, while reducing the synthetic-to-real distribution shift, which aligns with the synthetic experiments. Moreover, current inference-time interventions can augment the diversity of the generations at the expense of moving outside the support of real data. Among those interventions, prompt expansion, by deliberately using a pre-trained language model as a likelihood estimator, consistently achieves the highest performance in both image diversity and aesthetics, even higher than that of real data. Combining advanced guidance interventions with prompt expansion results in the most appealing utility trade-offs of synthetic data.
2025-12-31
International Conference on Learning Representations (Accept (Poster))
openreview.net
openreview.net
The Markovian Thinker
Milad Aghajohari
Kamran Chitsaz
Amirhossein Kazemnejad
Sarath Chandar
Alessandro Sordoni
Aaron Courville
Siva Reddy
Reasoning LLMs suffer from quadratic compute growth as their context length increases, making reinforcement learning with verifiable rewards… (voir plus) (RLVR) and test-time scaling prohibitively expensive. Prior work has tried to lighten the computational burden by shortening reasoning traces through pruning, summarization, or multi-stage training, but these methods remain bound to quadratic costs. We introduce Delethink, a thinking algorithm that realizes the Markovian Thinking Paradigm. Instead of producing one long monolithic reasoning trace, Delethink thinks in a sequence of chunks, the Delethink trace. Each chunk continues reasoning by referring only to a fixed number of prior tokens, which functions as a Markovian state sufficient for progressing reasoning, while deleting the rest. This preserves continuity without carrying the quadratic baggage. As a result, compute scales linearly and peak memory remains constant. In experiments, we show that Delethink can be applied directly to off-the-shelf reasoning models ranging from
2025-12-31
International Conference on Learning Representations (Accept (Poster))
openreview.net
openreview.net
On the sensitivity of SAR C- and L-band dual-polarized data for detection of early deforestation in the tropics
Africa I. Flores-Anderson
Jeffrey A. Cardille
Josef Kellndorfer
Franz J. Meyer
Pontus Olofsson
2025-12-31
Remote Sensing of Environment (publié)
doi.org
The Surprising Difficulty of Search in Model-Based Reinforcement Learning
Wei-Di Chang
Mikael Henaff
Brandon Amos
Gregory Dudek
Scott Fujimoto
This paper investigates search in model-based reinforcement learning (RL). Conventional wisdom holds that long-term predictions and compound… (voir plus)ing errors are the primary obstacles for model-based RL. We challenge this view, showing that search is not a plug-and-play replacement for a learned policy. Surprisingly, we find that search can harm performance even when the model is highly accurate. Instead, we show that mitigating distribution shift matters more than improving model or value function accuracy. Building on this insight, we identify key techniques for enabling effective search, achieving state-of-the-art performance across multiple popular benchmark domains.
2025-12-31
arXiv (prépublication)
doi.org
arxiv.org
Think Fast: Real-Time IoT Intrusion Reasoning Using IDS and LLMs at the Edge Gateway
Saeid Jamshidi
Omar Abdel Wahab
Rolando Herrero
Foutse Khomh
Martine Bellaiche
Samira Keivanpour
Negar Shahabi
Amin Nikanjam
Kawser Wazed Nafi
2025-12-31
IEEE Internet of Things Journal (publié)
doi.org
arxiv.org
Toward Faithful Explanations in Acoustic Anomaly Detection
Maab Elrashid
Anthony Deschênes
Yusuf Cem Sübakan
Mirco Ravanaelli
Rémi Georges
Michael Morin
Interpretability is essential for user trust in real-world anomaly detection applications. However, deep learning models, despite their stro… (voir plus)ng performance, often lack transparency. In this work, we study the interpretability of autoencoder-based models for audio anomaly detection, by comparing a standard autoencoder (AE) with a mask autoencoder (MAE) in terms of detection performance and interpretability. We applied several attribution methods, including error maps, saliency maps, SmoothGrad, Integrated Gradients, GradSHAP, and Grad-CAM. Although MAE shows a slightly lower detection, it consistently provides more faithful and temporally precise explanations, suggesting a better alignment with true anomalies. To assess the relevance of the regions highlighted by the explanation method, we propose a perturbation-based faithfulness metric that replaces them with their reconstructions to simulate normal input. Our findings, based on experiments in a real industrial scenario, highlight the importance of incorporating interpretability into anomaly detection pipelines and show that masked training improves explanation quality without compromising performance.
2025-12-31
arXiv (publié)
doi.org
arxiv.org
Towards All-Atom Foundation Models for Biomolecular Binding Affinity Prediction
Liang Shi
Zuobai Zhang
Huiyu Cai
Santiago Miret
Zhi Yang
Jian Tang
Biomolecular interactions play a critical role in biological processes. While recent breakthroughs like AlphaFold 3 have enabled accurate mo… (voir plus)deling of biomolecular complex structures, predicting binding affinity remains challenging mainly due to limited high-quality data. Recent methods are often specialized for specific types of biomolecular interactions, limiting their generalizability. In this work, we repurpose AlphaFold 3 for representation learning to predict binding affinity, a non-trivial task that requires shifting from generative structure prediction to encoding observed geometry, simplifying the heavily conditioned trunk module, and designing a framework to jointly capture sequence and structural information. To address these challenges, we introduce the **Atom-level Diffusion Transformer (ADiT)**, which takes sequence and structure as inputs, employs a unified tokenization scheme, integrates diffusion transformers, and removes dependencies on multiple sequence alignments and templates. We pre-train three ADiT variants on the PDB dataset with a denoising objective and evaluate them across protein-ligand, drug-target, protein-protein, and antibody-antigen interactions. The model achieves state-of-the-art or competitive performance across benchmarks, scales effectively with model size, and successfully identifies wet-lab validated affinity-enhancing antibody mutations, establishing a generalizable framework for biomolecular interactions. We plan to release the code upon acceptance.
2025-12-31
International Conference on Learning Representations (Accept (Poster))
openreview.net
openreview.net
Towards Learned Optimization Free Lunch
Abhinav Moudgil
Boris Knyazev
Eugene Belilovsky
Learned optimizers are powerful alternatives to hand-designed rules like Adam, yet they have seen limited practical adoption since they ofte… (voir plus)n fail to meta-generalize beyond their training distribution and incur high meta-training cost. For instance, prior work, VeLO, scaled meta-training to 4,000 TPU months (
2025-12-31
International Conference on Learning Representations (Accept (Poster))
openreview.net
openreview.net
Towards Multi-Brain Decoding in Autism: A Self-Supervised Learning Approach
Ghazaleh Ranjabaran
Quentin Moreau
Adrien Dubois
Guillaume Dumas
Abstract This study introduces a self-supervised learning (SSL) approach to hyperscanning electroencephalog… (voir plus)raphy (EEG) data, targeting the identification of autism spectrum condition (ASC) during social interactions. Hyperscanning enables simultaneous recording of neural activity across interacting individuals, offering a novel path for studying brain-to-brain synchrony in ASC. Leveraging a large-scale, single-brain EEG dataset for SSL pretraining, we developed a multi-brain classification model fine-tuned with hyperscanning data from dyadic interactions involving ASC and neurotypical participants. The SSL model demonstrated superior performance (78.13% accuracy) compared to supervised baselines and logistic regression using spectral EEG biomarkers. These results underscore the efficacy of SSL in addressing the challenges of limited labeled data, enhancing EEG-based diagnostic tools for ASC, and advancing research in social neuroscience.
2025-12-31
Neuroinformatics (publié)
doi.org
TRecViT: A Recurrent Video Transformer
Viorica Patraucean
Xu Owen He
Joseph Heyward
Chuhan Zhang
Mehdi S. M. Sajjadi
George-Cristian Muraru
Artem Zholus
Mahdi Karami
Ross Goroshin
Yutian Chen 0001
Simon Kayode Osindero
João Carreira
Razvan Pascanu
We propose a novel block for video modelling. It relies on a time-space-channel factorisation with dedicated blocks for each dimension: gate… (voir plus)d linear recurrent units (LRUs) perform information mixing over time, self-attention layers perform mixing over space, and MLPs over channels. The resulting architecture TRecViT performs well on sparse and dense tasks, trained in supervised or self-supervised regimes. Notably, our model is causal and outperforms or is on par with a pure attention model ViViT-L on large scale video datasets (SSv2, Kinetics400), while having
2025-12-31
Trans. Mach. Learn. Res. (publié)
doi.org
arxiv.org
Understanding and Exploiting Weight Update Sparsity for Communication-Efficient Distributed RL.
Erfan Miahi
Eugene Belilovsky
Reinforcement learning (RL) is a critical component for post-training large language models (LLMs). However, in bandwidth-constrained distri… (voir plus)buted RL, scalability is often bottlenecked by the synchronization of policy weights from trainers to inference workers, particularly over commodity networks or in decentralized settings. While recent studies suggest that RL updates modify only a small fraction of model parameters, these observations are typically based on coarse checkpoint differences. We present a systematic empirical study of weight-update sparsity at both step-level and multi-step granularities, examining its evolution across training dynamics, off-policy delay, and model scale. We find that update sparsity is consistently high, frequently exceeding 99% across practically relevant settings. Leveraging this structure, we propose PULSE (Patch Updates via Lossless Sparse Encoding), a simple yet highly efficient lossless weight synchronization method that transmits only the indices and values of modified parameters. PULSE is robust to transmission errors and avoids floating-point drift inherent in additive delta schemes. In bandwidth-constrained decentralized environments, our approach achieves over 100x (14 GB to ~108 MB) communication reduction while maintaining bit-identical training dynamics and performance compared to full weight synchronization. By exploiting this structure, PULSE enables decentralized RL training to approach centralized throughput, reducing the bandwidth required for weight synchronization from 20 Gbit/s to 0.2 Gbit/s to maintain high GPU utilization.
2025-12-31
arXiv (prépublication)
doi.org
arxiv.org
Unsupervised Classification of English Words Based on Phonological Information: Discovery of Germanic and Latinate Clusters
Takashi Morita
Timothy J. O’Donnell
Abstract Cross-linguistically, native words and loanwords follow different phonological rules. In English, for example, words of Germanic an… (voir plus)d Latinate origin exhibit different stress patterns, and a certain syntactic structure—double-object datives—is predominantly associated with Germanic verbs rather than Latinate verbs. From the perspective of language acquisition, however, such etymology-based generalizations raise learnability concerns, since the historical origins of words are presumably inaccessible information for general language learners. In this study, we present computational evidence indicating that the Germanic-Latinate distinction in the English lexicon is learnable from the phonotactic information of individual words. Specifically, we performed an unsupervised clustering on corpus-extracted words, and the resulting word clusters largely aligned with the etymological distinction. The model-discovered clusters also recovered various linguistic generalizations documented in the previous literature regarding the corresponding etymological classes. Moreover, our model also uncovered previously unrecognized features of the quasi-etymological clusters. Taken together with prior results from Japanese, our findings indicate that the proposed method provides a general, cross-linguistic approach to discovering etymological structure from phonotactic cues in the lexicon.
2025-12-31
Open Mind: Discoveries in Cognitive Science (publié)
doi.org
arxiv.org