Publications

Personalized inference for neurostimulation with meta-learning: a case study of vagus nerve stimulation
Ximeng Mao
Yao-Chuan Chang
Stavros Zanos
Guillaume Lajoie
2024-01-11
Journal of Neural Engineering (publié)
doi.org
DINOv2: Learning Robust Visual Features without Supervision
Maxime Oquab
Timothée Darcet
Théo Moutakanni
Huy V. Vo
Marc Szafraniec
Vasil Khalidov
Pierre Fernandez
Daniel HAZIZA
Francisco Massa
Alaaeldin El-Nouby
Mahmoud Assran
Nicolas Ballas
Wojciech Galuba
Russell Howes
Po-Yao Huang
Shang-Wen Li
Ishan Misra
Michael G. Rabbat
Vasu Sharma
Gabriel Synnaeve … (voir 8 de plus)
Hu Xu 0001
Huijiao Xu
Hu Xu
Herve Jegou
Julien Mairal
Patrick Labatut
Armand Joulin
Piotr Bojanowski
The recent breakthroughs in natural language processing for model pretraining on large quantities of data have opened the way for similar fo… (voir plus)undation models in computer vision. These models could greatly simplify the use of images in any system by producing all-purpose visual features, i.e., features that work across image distributions and tasks without finetuning. This work shows that existing pretraining methods, especially self-supervised methods, can produce such features if trained on enough curated data from diverse sources. We revisit existing approaches and combine different techniques to scale our pretraining in terms of data and model size. Most of the technical contributions aim at accelerating and stabilizing the training at scale. In terms of data, we propose an automatic pipeline to build a dedicated, diverse, and curated image dataset instead of uncurated data, as typically done in the self-supervised literature. In terms of models, we train a ViT model with 1B parameters and distill it into a series of smaller models that surpass the best available all-purpose features, OpenCLIP on most of the benchmarks at image and pixel levels.
2024-01-10
TMLR (accepté)
doi.org
openreview.net
A database of the healthy human spinal cord morphometry in the PAM50 template space
Jan Valosek
Sandrine Bédard
Miloš Keřkovský
Tomáš Rohan
Julien Cohen-Adad
Measures of spinal cord morphometry computed from magnetic resonance images serve as relevant prognostic biomarkers for a range of spinal co… (voir plus)rd pathologies, including traumatic and non-traumatic spinal cord injury and neurodegenerative diseases. However, interpreting these imaging biomarkers is difficult due to considerable intra- and inter-subject variability. Yet, there is no clear consensus on a normalization method that would help reduce this variability and more insights into the distribution of these morphometrics are needed. In this study, we computed a database of normative values for six commonly used measures of spinal cord morphometry: cross-sectional area, anteroposterior diameter, transverse diameter, compression ratio, eccentricity, and solidity. Normative values were computed from a large open-access dataset of healthy adult volunteers (N = 203) and were brought to the common space of the PAM50 spinal cord template using a newly proposed normalization method based on linear interpolation. Compared to traditional image-based registration, the proposed normalization approach does not involve image transformations and, therefore, does not introduce distortions of spinal cord anatomy. This is a crucial consideration in preserving the integrity of the spinal cord anatomy in conditions such as spinal cord injury. This new morphometric database allows researchers to normalize based on sex and age, thereby minimizing inter-subject variability associated with demographic and biological factors. The proposed methodology is open-source and accessible through the Spinal Cord Toolbox (SCT) v6.0 and higher.
2024-01-09
Imaging Neuroscience (publié)
doi.org
Nonparametric Partial Disentanglement via Mechanism Sparsity: Sparse Actions, Interventions and Sparse Temporal Dependencies
Sébastien Lachapelle
Pau Rodríguez
Yash Sharma
Katie Everett
Rémi Le Priol
Alexandre Lacoste
Simon Lacoste-Julien
2024-01-09
ArXiv (prépublication)
doi.org
arxiv.org
DyG2Vec: Efficient Representation Learning for Dynamic Graphs
Mohammad Alomrani
Mahdi Biparva
Yingxue Zhang
Mark J. Coates
Temporal graph neural networks have shown promising results in learning inductive representations by automatically extracting temporal patte… (voir plus)rns. However, previous works often rely on complex memory modules or inefficient random walk methods to construct temporal representations. To address these limitations, we present an efficient yet effective attention-based encoder that leverages temporal edge encodings and window-based subgraph sampling to generate task-agnostic embeddings. Moreover, we propose a joint-embedding architecture using non-contrastive SSL to learn rich temporal embeddings without labels. Experimental results on 7 benchmark datasets indicate that on average, our model outperforms SoTA baselines on the future link prediction task by 4.23% for the transductive setting and 3.30% for the inductive setting while only requiring 5-10x less training/inference time. Lastly, different aspects of the proposed framework are investigated through experimental analysis and ablation studies. The code is publicly available at https://github.com/huawei-noah/noah-research/tree/master/graph_atlas.
2024-01-07
TMLR (accepté)
openreview.net
openreview.net
CO emission predictions in municipal solid waste incineration based on reduced depth features and long short-term memory optimization
Runyu Zhang
Jian Tang
Heng Xia
Xiaotong Pan
Wen Yu
JunFei Qiao
2024-01-07
Neural computing & applications (Print) (publié)
doi.org
JaxPruner: A concise library for sparsity research
Joo Hyung Lee
Wonpyo Park
Nicole Elyse Mitchell
Jonathan Pilault
Johan Samir Obando Ceron
Han-Byul Kim
Namhoon Lee
Elias Frantar
Yun Long
Amir Yazdanbakhsh
Shivani Agrawal
Suvinay Subramanian
Xin Wang
Sheng-Chun Kao
Xingyao Zhang
Trevor Gale
Aart J.C. Bik
Woohyun Han
Milen Ferev
Zhonglin Han … (voir 5 de plus)
Hong-Seok Kim
Yann Dauphin
Gintare Karolina Dziugaite
Pablo Samuel Castro
Utku Evci
This paper introduces JaxPruner, an open-source JAX-based pruning and sparse training library for machine learning research. JaxPruner aims … (voir plus)to accelerate research on sparse neural networks by providing concise implementations of popular pruning and sparse training algorithms with minimal memory and latency overhead. Algorithms implemented in JaxPruner use a common API and work seamlessly with the popular optimization library Optax, which, in turn, enables easy integration with existing JAX based libraries. We demonstrate this ease of integration by providing examples in four different codebases: Scenic, t5x, Dopamine and FedJAX and provide baseline experiments on popular benchmarks.
2024-01-07
Conference on Parsimony and Learning (publié)
doi.org
proceedings.mlr.press
GABAergic inhibition shapes behavior and neural dynamics in human visual working memory
Jan Kujala
Carolina Ciumas
Julien Jung
Sandrine Bouvard
Françoise Lecaignard
Amélie Lothe
Romain Bouet
Philippe Ryvlin
Karim Jerbi CoCo Lab
Abstract Neuronal inhibition, primarily mediated by GABAergic neurotransmission, is crucial for brain development and healthy cognition. Gam… (voir plus)ma-aminobutyric acid concentration levels in sensory areas have been shown to correlate with hemodynamic and oscillatory neuronal responses. How these measures relate to one another during working memory, a higher-order cognitive process, is still poorly understood. We address this gap by collecting magnetoencephalography, functional magnetic resonance imaging, and Flumazenil positron emission tomography data within the same subject cohort using an n-back working-memory paradigm. By probing the relationship between GABAA receptor distribution, neural oscillations, and Blood Oxygen Level Dependent (BOLD) modulations, we found that GABAA receptor density in higher-order cortical areas predicted the reaction times on the working-memory task and correlated positively with the peak frequency of gamma power modulations and negatively with BOLD amplitude. These findings support and extend theories linking gamma oscillations and hemodynamic responses to gamma-aminobutyric acid neurotransmission and to the excitation-inhibition balance and cognitive performance in humans. Considering the small sample size of the study, future studies should test whether these findings also hold for other, larger cohorts as well as to examine in detail how the GABAergic system and neural fluctuations jointly support working-memory task performance.
2024-01-05
Cerebral Cortex (publié)
doi.org
Functional Labeled Optimal Partitioning
Jacob M. Kaufman
Alyssa J. Stenberg
Toby Dylan Hocking
2024-01-04
Journal of Computational And Graphical Statistics (publié)
doi.org
On the Stability of a non-hyperbolic nonlinear map with non-bounded set of non-isolated fixed points with applications to Machine Learning
Roberta Hansen
Matias Vera
Lautaro Estienne
LUCIANA FERRER
Pablo Piantanida
2024-01-04
ArXiv (prépublication)
arxiv.org
arxiv.org
Are LLMs Robust for Spoken Dialogues?
Seyed Mahed Mousavi
Gabriel Roccabruna
Simone Alghisi
Massimo Rizzoli
Mirco Ravanaelli
Giuseppe Riccardi
Large Pre-Trained Language Models have demonstrated state-of-the-art performance in different downstream tasks, including dialogue state tra… (voir plus)cking and end-to-end response generation. Nevertheless, most of the publicly available datasets and benchmarks on task-oriented dialogues focus on written conversations. Consequently, the robustness of the developed models to spoken interactions is unknown. In this work, we have evaluated the performance of LLMs for spoken task-oriented dialogues on the DSTC11 test sets. Due to the lack of proper spoken dialogue datasets, we have automatically transcribed a development set of spoken dialogues with a state-of-the-art ASR engine. We have characterized the ASR-error types and their distributions and simulated these errors in a large dataset of dialogues. We report the intrinsic (perplexity) and extrinsic (human evaluation) performance of fine-tuned GPT-2 and T5 models in two subtasks of response generation and dialogue state tracking, respectively. The results show that LLMs are not robust to spoken noise by default, however, fine-tuning/training such models on a proper dataset of spoken TODs can result in a more robust performance.
2024-01-03
ArXiv (prépublication)
doi.org
arxiv.org
A primer on the use of machine learning to distil knowledge from data in biological psychiatry.
Thomas P. Quinn
Jonathan L. Hess
Victoria S. Marshe
Michelle M. Barnett
Anne-Christin Hauschild
Malgorzata Maciukiewicz
Samar S. M. Elsheikh
Xiaoyu Men
Emanuel Schwarz
Yannis Trakadis
Michael S. Breen
Eric J. Barnett
Yanli Zhang-James
Mehmet Eren Ahsen
Han Cao
Junfang Chen
Jiahui Hou
Asif Salekin
Ping-I Lin
Kristin K. Nicodemus … (voir 7 de plus)
Andreas Meyer-Lindenberg
Isabelle Bichindaritz
Stephen V. Faraone
Murray J. Cairns
Gaurav Pandey
Daniel J. Müller
Stephen J. Glatt
2024-01-03
Molecular Psychiatry (publié)
doi.org