Publications

Cascaded Video Generation for Videos In-the-Wild
Lluis Castrejon
Nicolas Ballas
Aaron Courville
Videos can be created by first outlining a global view of the scene and then adding local details. Inspired by this idea we propose a cascad… (see more)ed model for video generation which follows a coarse to fine approach. First our model generates a low resolution video, establishing the global scene structure, which is then refined by subsequent cascade levels operating at larger resolutions. We train each cascade level sequentially on partial views of the videos, which reduces the computational complexity of our model and makes it scalable to high-resolution videos with many frames. We empirically validate our approach on UCF101 and Kinetics-600, for which our model is competitive with the state-of-the-art. We further demonstrate the scaling capabilities of our model and train a three-level model on the BDD100K dataset which generates 256x256 pixels videos with 48 frames.
2022-08-20
2022 26th International Conference on Pattern Recognition (ICPR) (published)
doi.org
arxiv.org
Privacy-aware compression for federated data analysis
Kamalika Chaudhuri
Chuan Guo
Michael G. Rabbat
Federated data analytics is a framework for distributed data analysis where a server compiles noisy responses from a group of distributed lo… (see more)w-bandwidth user devices to estimate aggregate statistics. Two major challenges in this framework are privacy, since user data is often sensitive, and compression, since the user devices have low network bandwidth. Prior work has addressed these challenges separately by combining standard compression algorithms with known privacy mechanisms. In this work, we take a holistic look at the problem and design a family of privacy-aware compression mechanisms that work for any given communication budget. We first propose a mechanism for transmitting a single real number that has optimal variance under certain conditions. We then show how to extend it to metric differential privacy for location privacy use-cases, as well as vectors, for application to federated learning. Our experiments illustrate that our mechanism can lead to better utility vs. compression trade-offs for the same privacy loss in a number of settings.
2022-08-16
Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence (published)
doi.org
proceedings.mlr.press
Probabilistic surrogate networks for simulators with unbounded randomness
Andreas Munk
Berend Zwartsenberg
Adam Ścibior
Atilim Güneş Baydin
Andrew Lawrence Stewart
Goran Fernlund
Anoush Poursartip
Frank N. Wood
We present a framework for automatically structuring and training fast, approximate, deep neural surrogates of stochastic simulators. Unlike… (see more) traditional approaches to surrogate modeling, our surrogates retain the interpretable structure and control flow of the reference simulator. Our surrogates target stochastic simulators where the number of random variables itself can be stochastic and potentially unbounded. Our framework further enables an automatic replacement of the reference simulator with the surrogate when undertaking amortized inference. The fidelity and speed of our surrogates allow for both faster stochastic simulation and accurate and substantially faster posterior inference. Using an illustrative yet non-trivial example we show our surrogates' ability to accurately model a probabilistic program with an unbounded number of random variables. We then proceed with an example that shows our surrogates are able to accurately model a complex structure like an unbounded stack in a program synthesis example. We further demonstrate how our surrogate modeling technique makes amortized inference in complex black-box simulators an order of magnitude faster. Specifically, we do simulator-based materials quality testing, inferring safety-critical latent internal temperature profiles of composite materials undergoing curing.
2022-08-16
Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence (published)
proceedings.mlr.press
proceedings.mlr.press
Stratifying the autistic phenotype using electrophysiological indices of social perception
Luke Mason
Carolin Moessnang
Christopher H. Chatham
Lindsay Ham
Julian Tillmann
Guillaume Dumas
Claire Ellis
Claire Leblond
Freddy Cliquet
Thomas Bourgeron
Christian Beckmann
Tony Charman
Beth Oakley
Tobias Banaschewski
Andreas Meyer-Lindenberg
Simon Baron-Cohen
Sven Bölte
Jan K. Buitelaar
Sarah Durston
Eva Loth … (see 7 more)
Bob Oranje
Antonio Persico
Flavio Dell’Acqua
Christine Ecker
Mark Johnson
Declan Murphy
Emily J. H. Jones
Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by difficulties in social communication, but also great heter… (see more)ogeneity. To offer individualized medicine approaches, we need to better target interventions by stratifying autistic people into subgroups with different biological profiles and/or prognoses. We sought to validate neural responses to faces as a potential stratification factor in ASD by measuring neural (electroencephalography) responses to faces (critical in social interaction) in N = 436 children and adults with and without ASD. The speed of early-stage face processing (N170 latency) was on average slower in ASD than in age-matched controls. In addition, N170 latency was associated with responses to faces in the fusiform gyrus, measured with functional magnetic resonance imaging, and polygenic scores for ASD. Within the ASD group, N170 latency predicted change in adaptive socialization skills over an 18-month follow-up period; data-driven clustering identified a subgroup with slower brain responses and poor social prognosis. Use of a distributional data-driven cutoff was associated with predicted improvements of power in simulated clinical trials targeting social functioning. Together, the data provide converging evidence for the utility of the N170 as a stratification factor to identify biologically and prognostically defined subgroups in ASD. Description N170 latency to faces relates to fusiform activity and ASD genetics, predicts social prognosis, and could improve power in clinical trials. Exploiting face processing in patients with ASD The heterogeneity observed in patients with autism spectrum disorder (ASD) highlights the need for better patient stratification methods. Here, Mason et al. evaluated the use of the speed of early-stage face processing (N170 latency) for patient stratification and prognosis in subjects with ASD and age-matched healthy individuals. N170 latency was slower in individuals with ASD and correlated with response to faces measured with fMRI and with polygenic risk score. Among subjects with ASD, the N170 values stratified patients according to socialization prognosis and improved power in a simulated clinical trial. The results suggest that including N170 evaluation in patient stratification might help the design and development of patient-specific therapies for ASD.
2022-08-16
Science Translational Medicine (published)
doi.org
Towards Painless Policy Optimization for Constrained MDPs
Arushi Jain
Sharan Vaswani
Reza Babanezhad Harikandeh
Csaba Szepesvari
Doina Precup
We study policy optimization in an infinite horizon, …
2022-08-16
Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence (published)
doi.org
proceedings.mlr.press
Accepted Tutorials at The Web Conference 2022
Riccardo Tommasini
Senjuti Basu Roy
Xuan Wang
Hongwei Wang
Heng Ji
Jiawei Han
Preslav Nakov
Giovanni Da San Martino
Firoj Alam
Markus Schedl
Elisabeth Lex
Akash Bharadwaj
Graham Cormode
Milan Dojchinovski
Jan Forberg
Johannes Frey
Pieter Bonte
Marco Balduini
Matteo Belcao
Emanuele Della Valle … (see 53 more)
Junliang Yu
Hongzhi Yin
Tong Chen
Haochen Liu
Yiqi Wang
Wenqi Fan
Xiaorui Liu
Jamell Dacon
Lingjuan Lye
Jiliang Tang
Aristides Gionis
Stefan Neumann
Bruno Ordozgoiti
Simon Razniewski
Hiba Arnaout
Shrestha Ghosh
Fabian Suchanek
Lingfei Wu
Yu Chen
Yunyao Li
Bang Liu
Filip Ilievski
Daniel Garijo
Hans Chalupsky
Pedro Szekely
Ilias Kanellos
Dimitris Sacharidis
Thanasis Vergoulis
Nurendra Choudhary
Nikhil Rao
Karthik Subbian
Srinivasan Sengamedu
Chandan K. Reddy
Friedhelm Victor
Bernhard Haslhofer
George Katsogiannis- Meimarakis
Georgia Koutrika
Shengmin Jin
Danai Koutra
Reza Zafarani
Yulia Tsvetkov
Vidhisha Balachandran
Sachin Kumar
Xiangyu Zhao
Bo Chen
Huifeng Guo
Yejing Wang
Ruiming Tang
Yang Zhang
Wenjie Wang
Peng Wu
Fuli Feng
Xiangnan He
This paper summarizes the content of the 20 tutorials that have been given at The Web Conference 2022: 85% of these tutorials are lecture st… (see more)yle, and 15% of these are hands on.
2022-08-15
Companion Proceedings of the Web Conference 2022 (published)
doi.org
Pandemic policy assessment by artificial intelligence
Sirui Song
Xue Liu
Ying Li
Yang Yu
2022-08-15
Scientific Reports (published)
doi.org
VisPaD: Visualization and Pattern Discovery for Fighting Human Trafficking
Pratheeksha Nair
Yifei Li
Catalina Vajiac
Andreas Olligschlaeger
Meng-Chieh Lee
Namyong Park
Duen Horng Chau
Christos Faloutsos
Reihaneh Rabbany
Chieh Lee
2022-08-15
Companion Proceedings of the Web Conference 2022 (published)
doi.org
Good Scientific Practice in MEEG Research: Progress and Perspectives
Guiomar Niso
Laurens R. Krol
Etienne Combrisson
A.-Sophie Dubarry
Madison A. Elliott
Clément François
Yseult Héjja-Brichard
Sophie K. Herbst
Karim Jerbi
Vanja Kovic
Katia Lehongre
Steven J. Luck
Manuel Mercier
John C. Mosher
Yuri G. Pavlov
Aina Puce
Antonio Schettino
Daniele Schön
Walter Sinnott-Armstrong
Bertille Somon … (see 6 more)
Anđela Šoškić
Suzy J. Styles
Roni Tibon
Martina G. Vilas
Marijn van Vliet
Maximilien Chaumon
Good scientific practice (GSP) refers to both explicit and implicit rules, recommendations, and guidelines that help scientists to produce w… (see more)ork that is of the highest quality at any given time, and to efficiently share that work with the community for further scrutiny or utilization. For experimental research using magneto- and electroencephalography (MEEG), GSP includes specific standards and guidelines for technical competence, which are periodically updated and adapted to new findings. However, GSP also needs to be regularly revisited in a broader light. At the LiveMEEG 2020 conference, a reflection on GSP was fostered that included explicitly documented guidelines and technical advances, but also emphasized intangible GSP: a general awareness of personal, organizational, and societal realities and how they can influence MEEG research. This article provides an extensive report on most of the LiveMEEG contributions and new literature, with the additional aim to synthesize ongoing cultural changes in GSP. It first covers GSP with respect to cognitive biases and logical fallacies, pre-registration as a tool to avoid those and other early pitfalls, and a number of resources to enable collaborative and reproducible research as a general approach to minimize misconceptions. Second, it covers GSP with respect to data acquisition, analysis, reporting, and sharing, including new tools and frameworks to support collaborative work. Finally, GSP is considered in light of ethical implications of MEEG research and the resulting responsibility that scientists have to engage with societal challenges. Considering among other things the benefits of peer review and open access at all stages, the need to coordinate larger international projects, the complexity of MEEG subject matter, and today’s prioritization of fairness, privacy, and the environment, we find that current GSP tends to favor collective and cooperative work, for both scientific and for societal reasons.
2022-08-14
NeuroImage (published)
doi.org
High Fidelity Visualization of What Your Self-Supervised Representation Knows About
Florian Bordes
Randall Balestriero
P Vincent
Discovering what is learned by neural networks remains a challenge. In self-supervised learning, classification is the most common task used… (see more) to evaluate how good a representation is. However, relying only on such downstream task can limit our understanding of what information is retained in the representation of a given input. In this work, we showcase the use of a Representation Conditional Diffusion Model (RCDM) to visualize in data space the representations learned by self-supervised models. The use of RCDM is motivated by its ability to generate high-quality samples -- on par with state-of-the-art generative models -- while ensuring that the representations of those samples are faithful i.e. close to the one used for conditioning. By using RCDM to analyze self-supervised models, we are able to clearly show visually that i) SSL (backbone) representation are not invariant to the data augmentations they were trained with -- thus debunking an often restated but mistaken belief; ii) SSL post-projector embeddings appear indeed invariant to these data augmentation, along with many other data symmetries; iii) SSL representations appear more robust to small adversarial perturbation of their inputs than representations trained in a supervised manner; and iv) that SSL-trained representations exhibit an inherent structure that can be explored thanks to RCDM visualization and enables image manipulation.
2022-08-14
TMLR (accepted)
openreview.net
openreview.net
Automatic Phenotyping by a Seed-guided Topic Model.
Ziyang Song
Yuanyi Hu
Aman Verma
David L. Buckeridge
Yue Li
Electronic health records (EHRs) provide rich clinical information and the opportunities to extract epidemiological patterns to understand a… (see more)nd predict patient disease risks with suitable machine learning methods such as topic models. However, existing topic models do not generate identifiable topics each predicting a unique phenotype. One promising direction is to use known phenotype concepts to guide topic inference. We present a seed-guided Bayesian topic model called MixEHR-Seed with 3 contributions: (1) for each phenotype, we infer a dual-form of topic distribution: a seed-topic distribution over a small set of key EHR codes and a regular topic distribution over the entire EHR vocabulary; (2) we model age-dependent disease progression as Markovian dynamic topic priors; (3) we infer seed-guided multi-modal topics over distinct EHR data types. For inference, we developed a variational inference algorithm. Using MixEHR-Seed, we inferred 1569 PheCode-guided phenotype topics from an EHR database in Quebec, Canada covering 1.3 million patients for up to 20-year follow-up with 122 million records for 8539 and 1126 unique diagnostic and drug codes, respectively. We observed (1) accurate phenotype prediction by the guided topics, (2) clinically relevant PheCode-guided disease topics, (3) meaningful age-dependent disease prevalence. Source code is available at GitHub: https://github.com/li-lab-mcgill/MixEHR-Seed.
2022-08-13
Knowledge Discovery and Data Mining (published)
doi.org
TITRATED: Learned Human Driving Behavior without Infractions via Amortized Inference
Vasileios Lioutas
Adam Ścibior
Frank N. Wood
2022-08-11
TMLR (accepted)
openreview.net
openreview.net