Publications

Predicting Infectiousness for Proactive Contact Tracing
Yoshua Bengio
Prateek Gupta
Tegan Maharaj
Nasim Rahaman
Martin Weiss
Tristan Deleu
Eilif Muller
Meng Qu
Victor Schmidt
Pierre-Luc St-Charles
Hannah Alsdurf
Olexa Bilanuik
David Buckeridge
Gaétan Marceau Caron
Pierre-Luc Carrier
Joumana Ghosn
Satya Ortiz-Gagné
Chris Pal
Irina Rish
Bernhard Schölkopf … (see 3 more)
Abhinav Sharma
Jian Tang
Andrew Williams
The COVID-19 pandemic has spread rapidly worldwide, overwhelming manual contact tracing in many countries and resulting in widespread lockdo… (see more)wns for emergency containment. Large-scale digital contact tracing (DCT) has emerged as a potential solution to resume economic and social activity while minimizing spread of the virus. Various DCT methods have been proposed, each making trade-offs between privacy, mobility restrictions, and public health. The most common approach, binary contact tracing (BCT), models infection as a binary event, informed only by an individual's test results, with corresponding binary recommendations that either all or none of the individual's contacts quarantine. BCT ignores the inherent uncertainty in contacts and the infection process, which could be used to tailor messaging to high-risk individuals, and prompt proactive testing or earlier warnings. It also does not make use of observations such as symptoms or pre-existing medical conditions, which could be used to make more accurate infectiousness predictions. In this paper, we use a recently-proposed COVID-19 epidemiological simulator to develop and test methods that can be deployed to a smartphone to locally and proactively predict an individual's infectiousness (risk of infecting others) based on their contact history and other information, while respecting strong privacy constraints. Predictions are used to provide personalized recommendations to the individual via an app, as well as to send anonymized messages to the individual's contacts, who use this information to better predict their own infectiousness, an approach we call proactive contact tracing (PCT). We find a deep-learning based PCT method which improves over BCT for equivalent average mobility, suggesting PCT could help in safe re-opening and second-wave prevention.
2021-05-02
International Conference on Learning Representations (Spotlight)
doi.org
openreview.net
Recurrent Independent Mechanisms
Anirudh Goyal
Alex Lamb
Jordan Hoffmann
Shagun Sodhani
Sergey Levine
Yoshua Bengio
Bernhard Schölkopf
Learning modular structures which reflect the dynamics of the environment can lead to better generalization and robustness to changes which … (see more)only affect a few of the underlying causes. We propose Recurrent Independent Mechanisms (RIMs), a new recurrent architecture in which multiple groups of recurrent cells operate with nearly independent transition dynamics, communicate only sparingly through the bottleneck of attention, and are only updated at time steps where they are most relevant. We show that this leads to specialization amongst the RIMs, which in turn allows for dramatically improved generalization on tasks where some factors of variation differ systematically between training and evaluation.
2021-05-02
International Conference on Learning Representations (Spotlight)
doi.org
openreview.net
Reinforcement Learning with Random Delays
Yann Bouteiller
Simon Ramstedt
Giovanni Beltrame
Christopher Pal
Jonathan Binas
Action and observation delays commonly occur in many Reinforcement Learning applications, such as remote control scenarios. We study the ana… (see more)tomy of randomly delayed environments, and show that partially resampling trajectory fragments in hindsight allows for off-policy multi-step value estimation. We apply this principle to derive Delay-Correcting Actor-Critic (DCAC), an algorithm based on Soft Actor-Critic with significantly better performance in environments with delays. This is shown theoretically and also demonstrated practically on a delay-augmented version of the MuJoCo continuous control benchmark.
2021-05-02
International Conference on Learning Representations (Poster)
doi.org
openreview.net
RNNLogic: Learning Logic Rules for Reasoning on Knowledge Graphs
Meng Qu
Junkun Chen
Louis-Pascal Xhonneux
Yoshua Bengio
Jian Tang
This paper studies learning logic rules for reasoning on knowledge graphs. Logic rules provide interpretable explanations when used for pred… (see more)iction as well as being able to generalize to other tasks, and hence are critical to learn. Existing methods either suffer from the problem of searching in a large search space (e.g., neural logic programming) or ineffective optimization due to sparse rewards (e.g., techniques based on reinforcement learning). To address these limitations, this paper proposes a probabilistic model called RNNLogic. RNNLogic treats logic rules as a latent variable, and simultaneously trains a rule generator as well as a reasoning predictor with logic rules. We develop an EM-based algorithm for optimization. In each iteration, the reasoning predictor is first updated to explore some generated logic rules for reasoning. Then in the E-step, we select a set of high-quality rules from all generated rules with both the rule generator and reasoning predictor via posterior inference; and in the M-step, the rule generator is updated with the rules selected in the E-step. Experiments on four datasets prove the effectiveness of RNNLogic.
2021-05-02
International Conference on Learning Representations (Poster)
doi.org
openreview.net
Saliency Is a Possible Red Herring When Diagnosing Poor Generalization
Joseph Viviano
Becks Simpson
Francis Dutil
Yoshua Bengio
Joseph Paul Cohen
Poor generalization is one symptom of models that learn to predict target variables using spuriously-correlated image features present only … (see more)in the training distribution instead of the true image features that denote a class. It is often thought that this can be diagnosed visually using attribution (aka saliency) maps. We study if this assumption is correct. In some prediction tasks, such as for medical images, one may have some images with masks drawn by a human expert, indicating a region of the image containing relevant information to make the prediction. We study multiple methods that take advantage of such auxiliary labels, by training networks to ignore distracting features which may be found outside of the region of interest. This mask information is only used during training and has an impact on generalization accuracy depending on the severity of the shift between the training and test distributions. Surprisingly, while these methods improve generalization performance in the presence of a covariate shift, there is no strong correspondence between the correction of attribution towards the features a human expert has labelled as important and generalization performance. These results suggest that the root cause of poor generalization may not always be spatially defined, and raise questions about the utility of masks as "attribution priors" as well as saliency maps for explainable predictions.
2021-05-02
International Conference on Learning Representations (Poster)
doi.org
openreview.net
Saliency Is a Possible Red Herring When Diagnosing Poor Generalization
Joseph Viviano
Becks Simpson
Francis Dutil
Yoshua Bengio
Joseph Paul Cohen
Poor generalization is one symptom of models that learn to predict target variables using spuriously-correlated image features present only … (see more)in the training distribution instead of the true image features that denote a class. It is often thought that this can be diagnosed visually using attribution (aka saliency) maps. We study if this assumption is correct. In some prediction tasks, such as for medical images, one may have some images with masks drawn by a human expert, indicating a region of the image containing relevant information to make the prediction. We study multiple methods that take advantage of such auxiliary labels, by training networks to ignore distracting features which may be found outside of the region of interest. This mask information is only used during training and has an impact on generalization accuracy depending on the severity of the shift between the training and test distributions. Surprisingly, while these methods improve generalization performance in the presence of a covariate shift, there is no strong correspondence between the correction of attribution towards the features a human expert has labelled as important and generalization performance. These results suggest that the root cause of poor generalization may not always be spatially defined, and raise questions about the utility of masks as "attribution priors" as well as saliency maps for explainable predictions.
2021-05-02
International Conference on Learning Representations (Poster)
doi.org
openreview.net
Spatially Structured Recurrent Modules
Nasim Rahaman
Anirudh Goyal
Muhammad Waleed Gondal
Manuel Wuthrich
Stefan Bauer
Yash Sharma
Yoshua Bengio
Bernhard Schölkopf
Capturing the structure of a data-generating process by means of appropriate inductive biases can help in learning models that generalise we… (see more)ll and are robust to changes in the input distribution. While methods that harness spatial and temporal structures find broad application, recent work has demonstrated the potential of models that leverage sparse and modular structure using an ensemble of sparingly interacting modules. In this work, we take a step towards dynamic models that are capable of simultaneously exploiting both modular and spatiotemporal structures. To this end, we model the dynamical system as a collection of autonomous but sparsely interacting sub-systems that interact according to a learned topology which is informed by the spatial structure of the underlying system. This gives rise to a class of models that are well suited for capturing the dynamics of systems that only offer local views into their state, along with corresponding spatial locations of those views. On the tasks of video prediction from cropped frames and multi-agent world modelling from partial observations in the challenging Starcraft2 domain, we find our models to be more robust to the number of available views and better capable of generalisation to novel tasks without additional training than strong baselines that perform equally well or better on the training distribution.
2021-05-02
International Conference on Learning Representations (Poster)
openreview.net
openreview.net
Systematic Generalisation with Group Invariant Predictions
Faruk Ahmed
Yoshua Bengio
Harm van Seijen
Aaron Courville
2021-05-02
International Conference on Learning Representations (Spotlight)
openreview.net
openreview.net
TDprop: Does Adaptive Optimization With Jacobi Preconditioning Help Temporal Difference Learning?
Joshua Romoff
Peter Henderson
David Kanaa
Emmanuel Bengio
Ahmed Touati
Pierre-Luc Bacon
Joelle Pineau
We investigate whether Jacobi preconditioning, accounting for the bootstrap term in temporal difference (TD) learning, can help boost perfor… (see more)mance of adaptive optimizers. Our method, TDprop, computes a per-parameter learning rate based on the diagonal preconditioning of the TD update rule. We show how this can be used in both n-step returns and TD(λ). Our theoretical findings demonstrate that including this additional preconditioning information is comparable to normal semi-gradient TD if the optimal learning rate is found for both via a hyperparameter search. This matches our experimental results. In Deep RL experiments using Expected SARSA, TDprop meets or exceeds the performance of Adam in all tested games under near-optimal learning rates, but a well-tuned SGD can yield similar performance in most settings. Our findings suggest that Jacobi preconditioning may improve upon Adam in Deep RL, but despite incorporating additional information from the TD bootstrap term, may not always be better than SGD. Moreover, they suggest that more theoretical investigations are needed to understand adaptive optimizers under optimal hyperparameter regimes in TD learning: simpler methods may, surprisingly, be theoretically comparable after a hyperparameter search.
2021-05-02
Adaptive Agents and Multi-Agent Systems (published)
doi.org
Associations Between Relative Morning Blood Pressure, Cerebral Blood Flow, and Memory in Older Adults Treated and Controlled for Hypertension
Adrián Noriega de la Colina
Atef Badji
Marie-Christine Robitaille-Grou
Christine Gagnon
Tommy Boshkovski
Maxime Lamarre-Cliche
Sven Joubert
Claudine J. Gauthier
Louis Bherer
Julien Cohen-Adad
Hélène Girouard
Supplemental Digital Content is available in the text. Hypertension, elevated morning blood pressure (BP) surges, and circadian BP variabili… (see more)ty constitute risk factors for cerebrovascular events. Nevertheless, while evidence indicates that hypertension is associated with cognitive dysfunctions, the link between BP variability and cognitive performance during aging is not clear. The purpose of this study is to determine the interaction between relative morning BP, cerebral blood flow (CBF) levels, and cognitive performance in hypertensive older adults with controlled BP under antihypertensive treatment. Eighty-four participants aged between 60 and 75 years old were separated into normotensive (n=51) and hypertensive (n=33) groups and underwent 24-hour ambulatory BP monitoring. They were also examined for CBF in the gray matter (CBF-GM) by magnetic resonance imaging and 5 cognitive domains: global cognition, working memory, episodic memory, processing speed, and executive functions. There was no difference in cognitive performance and CBF between normotensive and controlled hypertensive participants. Through a sensitivity analysis, we identified that, among relative morning BP variables, the best fit for CBF values in this cohort was the morning-evening difference in BP. The relative morning BP was negatively associated with CBF-GM in these hypertensive older adults only. In turn, CBF-GM levels were negatively associated with working and episodic memory scores in hypertensive older adults. This is the first extended study demonstrating an association between high relative morning BP and lower levels of CBF-GM, including the further impact of CBF-GM levels on the cognitive performance of specific domains in a community-based cohort of older adults with hypertension.
2021-04-30
Hypertension (published)
doi.org
Toward Causal Representation Learning
Bernhard Schölkopf
Francesco Locatello
Stefan Bauer
Nan Rosemary Ke
Nal Kalchbrenner
Anirudh Goyal
Yoshua Bengio
The two fields of machine learning and graphical causality arose and are developed separately. However, there is, now, cross-pollination and… (see more) increasing interest in both fields to benefit from the advances of the other. In this article, we review fundamental concepts of causal inference and relate them to crucial open problems of machine learning, including transfer and generalization, thereby assaying how causality can contribute to modern machine learning research. This also applies in the opposite direction: we note that most work in causality starts from the premise that the causal variables are given. A central problem for AI and causality is, thus, causal representation learning, that is, the discovery of high-level causal variables from low-level observations. Finally, we delineate some implications of causality for machine learning and propose key research areas at the intersection of both communities.
2021-04-30
ETH Zurich (published)
doi.org
Deep Generative Models for Galaxy Image Simulations
François Lanusse
Rachel Mandelbaum
Siamak Ravanbakhsh
Chun-Liang Li
Peter Freeman
Barnabás Póczos
Image simulations are essential tools for preparing and validating the analysis of current and future wide-field optical surveys. However, t… (see more)he galaxy models used as the basis for these simulations are typically limited to simple parametric light profiles, or use a fairly limited amount of available space-based data. In this work, we propose a methodology based on Deep Generative Models to create complex models of galaxy morphologies that may meet the image simulation needs of upcoming surveys. We address the technical challenges associated with learning this morphology model from noisy and PSF-convolved images by building a hybrid Deep Learning/physical Bayesian hierarchical model for observed images, explicitly accounting for the Point Spread Function and noise properties. The generative model is further made conditional on physical galaxy parameters, to allow for sampling new light profiles from specific galaxy populations. We demonstrate our ability to train and sample from such a model on galaxy postage stamps from the HST/ACS COSMOS survey, and validate the quality of the model using a range of second- and higher-order morphology statistics. Using this set of statistics, we demonstrate significantly more realistic morphologies using these deep generative models compared to conventional parametric models. To help make these generative models practical tools for the community, we introduce GalSim-Hub, a community-driven repository of generative models, and a framework for incorporating generative models within the GalSim image simulation software.
2021-04-29
Monthly Notices of the Royal Astronomical Society (published)
doi.org
arxiv.org