Publications

Multi-Agent Reinforcement Learning for Fast-Timescale Demand Response of Residential Loads
Vincent Mai
Philippe Maisonneuve
Tianyu Zhang
Hadi Nekoei
Liam Paull
Antoine Lesage-Landry
To integrate high amounts of renewable energy resources, electrical power grids must be able to cope with high amplitude, fast timescale var… (voir plus)iations in power generation. Frequency regulation through demand response has the potential to coordinate temporally flexible loads, such as air conditioners, to counteract these variations. Existing approaches for discrete control with dynamic constraints struggle to provide satisfactory performance for fast timescale action selection with hundreds of agents. We propose a decentralized agent trained with multi-agent proximal policy optimization with localized communication. We explore two communication frameworks: hand-engineered, or learned through targeted multi-agent communication. The resulting policies perform well and robustly for frequency regulation, and scale seamlessly to arbitrary numbers of houses for constant processing times.
2023-11-27
Machine Learning (inconnu)
doi.org
arxiv.org
Room-temperature correlated states in twisted bilayer MoS$_2$
Fanfan Wu
Qiaoling Xu
Qinqin Wang
Yanbang Chu
Li Li
Jian Tang
Jieying Liu
Jinpeng Tian
Yiru Ji
Le Liu
Yalong Yuan
Zhiheng Huang
Jiaojiao Zhao
Xiaozhou Zan
Kenji Watanabe
Takashi Taniguchi
Dongxia Shi
Gangxu Gu
Yang Xu
Lede Xian … (voir 3 de plus)
Wei Yang
Luojun Du
Guangyu Zhang
2023-11-27
ArXiv (prépublication)
arxiv.org
arxiv.org
Symmetry Breaking and Equivariant Neural Networks
Sékou-Oumar Kaba
Siamak Ravanbakhsh
Using symmetry as an inductive bias in deep learning has been proven to be a principled approach for sample-efficient model design. However,… (voir plus) the relationship between symmetry and the imperative for equivariance in neural networks is not always obvious. Here, we analyze a key limitation that arises in equivariant functions: their incapacity to break symmetry at the level of individual data samples. In response, we introduce a novel notion of 'relaxed equivariance' that circumvents this limitation. We further demonstrate how to incorporate this relaxation into equivariant multilayer perceptrons (E-MLPs), offering an alternative to the noise-injection method. The relevance of symmetry breaking is then discussed in various application domains: physics, graph representation learning, combinatorial optimization and equivariant decoding.
2023-11-27
NeurIPS.cc/2023/Workshop/NeurReps (présentation orale)
doi.org
openreview.net
On the Information Geometry of Vision Transformers
Sonia Joseph
Kumar Krishna Agrawal
Arna Ghosh
Blake Aaron Richards
2023-11-27
NeurIPS.cc/2023/Workshop/NeurReps (poster)
openreview.net
openreview.net
On the Varied Faces of Overparameterization in Supervised and Self-Supervised Learning
Matteo Gamba
Arna Ghosh
Kumar Krishna Agrawal
Blake Aaron Richards
Agrawal
Hossein Azizpour
Mårten Björkman
The quality of the representations learned by neural networks depends on several factors, including the loss function, learning algorithm, a… (voir plus)nd model architecture. In this work, we use information geometric measures to assess the representation quality in a principled manner. We demonstrate that the sensitivity of learned representations to input perturbations, measured by the spectral norm of the feature Jacobian, provides valuable information about downstream generalization. On the other hand, measuring the coefficient of spectral decay observed in the eigenspectrum of feature covariance provides insights into the global representation geometry. First, we empirically establish an equivalence between these notions of representation quality and show that they are inversely correlated. Second, our analysis reveals the varying roles that overparameterization plays in improving generalization. Unlike supervised learning, we observe that increasing model width leads to higher discriminability and less smoothness in the self-supervised regime. Furthermore, we report that there is no observable double descent phenomenon in SSL with non-contrastive objectives for commonly used parameterization regimes, which opens up new opportunities for tight asymptotic analysis. Taken together, our results provide a loss-aware characterization of the different role of overparameterization in supervised and self-supervised learning.
2023-11-27
NeurIPS.cc/2023/Workshop/NeurReps (poster)
openreview.net
openreview.net
1351. Predictors of Loss of Infectivity Among Healthcare Workers with Primary and Recurrent SARS-CoV-2 infection: An Observational Cohort Study
Stefka Dzieciolowska
Yves Longtin
Hugues Charest
Tonya Roy
Judith Fafard
Inès Levade
Jean Longtin
Leighanne Parkes
Jasmin Villeneuve
Patrice Savard
J. Corbeil
Gaston De Serres
Abstract Background Factors associated with loss of infectivity in healthcare workers (HCWs) with COVID-19 are poorly understood. Understand… (voir plus)ing predictive factors could help optimize return-to-work criteria and minimize absenteeism. Methods Prospective observational cohort study of HCWs with COVID-19 conducted between Feb 20 2022 and March 6 2023 in 20 institutions in Montreal, Canada, with clinical/laboratory follow-up on Day 5, 7 and 10 of infection. Infectivity was determined by viral culture (Vero E6 cells) on nasopharyngeal swabs. Predictors of loss of infectivity were investigated by univariate and multivariate logistic regression. Results Overall, 121 participants (79.3% female, mean age 40 years) were recruited. Most (n=107, 88.4%) had received ≥3 vaccines and 20 (16.5%) had a history of prior COVID-19. The proportion of HCWs with a positive viral culture decreased from 71.9% on day 5 of infection to 18.2% on day 10. The proportion of HCWs with a positive RT-PCR decreased from 93.3% (112/120) on day 5 (median Ct value, 23.4 [IQR, 20.6-27.9]) to 61.2% (74/120) on day 10 (median Ct value, 32.5 [IQR, 28.5 to undetectable]). Rapid antigen detection test (RADT) positivity decreased from 81.5% on day 5 to 34.2% on day 10. Participants with recurrent COVID-19 had lower likelihood of infectivity at each visit (OR on day 5, 0.14; 95% CI 0.05-0.40; p 0.001; OR on day 7, 0.04; 95% CI, 0.01-0.33; p=0.003) and none were infective on day 10 (p=0.02). At each visit, recurrent cases had higher median RT-PCR Ct values than primary infections (p 0.03) and were more likely to have a negative RADT result (p 0.01). By multivariate analysis, ongoing infectivity was associated with a RT-PCR Ct value 23 (adjusted OR [aOR] on day 5, 22.75; p 0.001; aOR on Day 7, 182.30; p 0.001; and aOR on Day 10; 24.71; p=0.02). A history of previous COVID-19 was associated with a lower probability of infectivity on Day 5 (aOR, 0.005; p=0.003). By contrast, symptom improvement (including fever) and RADT result were not independent predictors of loss of infectivity. Conclusion A lower RT-PCR Ct value is associated with ongoing infectivity, whereas COVID-19 reinfection is a predictor of loss of infectivity. These findings could help optimize return-to-work algorithms. Disclosures All Authors: No reported disclosures
2023-11-26
Open Forum Infectious Diseases (publié)
doi.org
Author Correction: 30×30 biodiversity gains rely on national coordination
Isaac Eckert
Andrea Brown
Dominique Caron
Federico Riva
Laura J. Pollock
2023-11-26
Nature Communications (publié)
doi.org
Exploring the multidimensional nature of repetitive and restricted behaviors and interests (RRBI) in autism: neuroanatomical correlates and clinical implications
Aline Lefebvre
Nicolas Traut
Amandine Pedoux
Anna Maruani
Anita Beggiato
Monique Elmaleh
David Germanaud
Anouck Amestoy
Myriam Ly‐Le Moal
Christopher H. Chatham
Lorraine Murtagh
Manuel Bouvard
Marianne Alisson
Marion Leboyer
Thomas Bourgeron
Roberto Toro
Guillaume Dumas
Clara A. Moreau
Richard Delorme
2023-11-26
Molecular Autism (publié)
doi.org
scGeneRythm: Using Neural Networks and Fourier Transformation to Cluster Genes by Time-Frequency Patterns in Single-Cell Data
Yiming Jia
Hao Wu
Jun Ding
2023-11-26
bioRxiv (prépublication)
doi.org
The search for the lost attractor
Mario Pasquato
Syphax Haddad
Pierfrancesco Di Cintio
Alexandre Adam
Pablo Lemos
Noé Dia
Mircea Petrache
Ugo Niccolò Di Carlo
Alessandro Alberto Trani
Laurence Perreault-Levasseur
Yashar Hezaveh
2023-11-26
arXiv (prépublication)
doi.org
arxiv.org
Hybrid Scattering Transform - Long Short-Term Memory Networks for Intrapartum Fetal Heart Rate Classification
"Derek Kweku DEGBEDZUI
Michael W Kuzniewicz
Marie-Coralie Cornet
Yvonne Wu
Heather Forquer
Lawrence Gerstley
Emily F. Hamilton
Doina Precup
P. Warrick
Robert E. Kearney
This study assessed the early detection of the increased risk of hypoxic ischemic encephalopathy using raw fetal heart rate and its transfor… (voir plus)mation with scattering transform and a long short-term memory recurrent neural network. There was no significant difference between the two approaches. However, the use of scattering transform produced lower computational demands. Considering scalability to the large data in our database and computational efficiency, the experiments involving scattering transform coefficients will be selected to conduct subsequent experiments. Future works will address the limitations of this study, including the low model performance.
2023-11-25
2023 Computing in Cardiology Conference (CinC) (publié)
doi.org
Propositional Logics for the Lawvere Quantale
Giorgio Bacci
Radu Mardare
Prakash Panangaden
Gordon Plotkin
2023-11-22
Electronic Notes in Theoretical Informatics and Computer Science (publié)
doi.org
arxiv.org