Publications

Options of Interest: Temporal Abstraction with Interest Functions
Khimya Khetarpal
Martin Klissarov
Maxime Chevalier-Boisvert
Pierre-Luc Bacon
Doina Precup
Temporal abstraction refers to the ability of an agent to use behaviours of controllers which act for a limited, variable amount of time. Th… (see more)e options framework describes such behaviours as consisting of a subset of states in which they can initiate, an internal policy and a stochastic termination condition. However, much of the subsequent work on option discovery has ignored the initiation set, because of difficulty in learning it from data. We provide a generalization of initiation sets suitable for general function approximation, by defining an interest function associated with an option. We derive a gradient-based learning algorithm for interest functions, leading to a new interest-option-critic architecture. We investigate how interest functions can be leveraged to learn interpretable and reusable temporal abstractions. We demonstrate the efficacy of the proposed approach through quantitative and qualitative results, in both discrete and continuous environments.
2020-04-02
Proceedings of the AAAI Conference on Artificial Intelligence (published)
doi.org
arxiv.org
Revision in Continuous Space: Unsupervised Text Style Transfer without Adversarial Learning
Dayiheng Liu
Jie Fu
Yidan Zhang
Christopher Pal
Jiancheng Lv
Typical methods for unsupervised text style transfer often rely on two key ingredients: 1) seeking the explicit disentanglement of the conte… (see more)nt and the attributes, and 2) troublesome adversarial learning. In this paper, we show that neither of these components is indispensable. We propose a new framework that utilizes the gradients to revise the sentence in a continuous space during inference to achieve text style transfer. Our method consists of three key components: a variational auto-encoder (VAE), some attribute predictors (one for each attribute), and a content predictor. The VAE and the two types of predictors enable us to perform gradient-based optimization in the continuous space, which is mapped from sentences in a discrete space, to find the representation of a target sentence with the desired attributes and preserved content. Moreover, the proposed method naturally has the ability to simultaneously manipulate multiple fine-grained attributes, such as sentence length and the presence of specific words, when performing text style transfer tasks. Compared with previous adversarial learning based methods, the proposed method is more interpretable, controllable and easier to train. Extensive experimental studies on three popular text style transfer tasks show that the proposed method significantly outperforms five state-of-the-art methods.
2020-04-02
Proceedings of the AAAI Conference on Artificial Intelligence (published)
doi.org
arxiv.org
Learning to cooperate: Emergent communication in multi-agent navigation
Ivana Kaji'c
Eser Aygün
Doina Precup
Emergent communication in artificial agents has been studied to understand language evolution, as well as to develop artificial systems that… (see more) learn to communicate with humans. We show that agents performing a cooperative navigation task in various gridworld environments learn an interpretable communication protocol that enables them to efficiently, and in many cases, optimally, solve the task. An analysis of the agents' policies reveals that emergent signals spatially cluster the state space, with signals referring to specific locations and spatial directions such as "left", "up", or "upper left room". Using populations of agents, we show that the emergent protocol has basic compositional structure, thus exhibiting a core property of natural language.
2020-04-01
ArXiv (preprint)
arxiv.org
arxiv.org
GitHub Repositories with Links to Academic Papers: Open Access, Traceability, and Evolution
Supatsara Wattanakriengkrai
Bodin Chinthanet
Hideaki Hata
Raula Gaikovina Kula
Christoph Treude
Jin L.C. Guo
Ken-ichi Matsumoto
Traceability between published scientific breakthroughs and their implementation is essential, especially in the case of Open Source Softwar… (see more)e implements bleeding edge science into its code. However, aligning the link between GitHub repositories and academic papers can prove difficult, and the link impact remains unknown. This paper investigates the role of academic paper references contained in these repositories. We conducted a large-scale study of 20 thousand GitHub repositories to establish prevalence of references to academic papers. We use a mixed-methods approach to identify Open Access (OA), traceability and evolutionary aspects of the links. Although referencing a paper is not typical, we find that a vast majority of referenced academic papers are OA. In terms of traceability, our analysis revealed that machine learning is the most prevalent topic of repositories. These repositories tend to be affiliated with academic communities. More than half of the papers do not link back to any repository. A case study of referenced arXiv paper shows that most of these papers are high-impact and influential and do align with academia, referenced by repositories written in different programming languages. From the evolutionary aspect, we find very few changes of papers being referenced and links to them.
2020-03-31
arXiv.org (preprint)
dblp.uni-trier.de
Pruning for efficient hardware implementations of deep neural networks
Ghouthi Boukli hacene
Vincent Gripon
Matthieu Arzel
Nicolas Farrugia
Yoshua Bengio
2020-03-31
(published)
www.semanticscholar.org
Both New and Chronic Potentially Inappropriate Medications Continued at Hospital Discharge Are Associated With Increased Risk of Adverse Events
Daniala L. Weir
Todd C. Lee
Emily G. McDonald
Aude Motulsky
Michal Abrahamowicz
Steven Morgan
David Buckeridge
Robyn Tamblyn
Admission to hospital provides the opportunity to review patient medications; however, the extent to which the safety of drug regimens chang… (see more)es after hospitalization is unclear. To estimate the number of potentially inappropriate medications (PIMs) prescribed to patients at hospital discharge and their association with the risk of adverse events 30 days after discharge. Prospective cohort study. Tertiary care hospitals within the McGill University Health Centre Network in Montreal, Quebec, Canada. Patients from internal medicine, cardiac, and thoracic surgery, aged 65 years and older, admitted between October 2014 and November 2016. Abstracted chart data were linked to provincial health databases. PIMs were identified using AGS (American Geriatrics Society) Beers Criteria®, STOPP, and Choosing Wisely statements. Multivariable logistic regression and Cox models were used to assess the association between PIMs and adverse events. Of 2,402 included patients, 1,381 (57%) were male; median age was 76 years (interquartile range [IQR] = 70‐82 years); and eight discharge medications were prescribed (IQR = 2‐8). A total of 1,576 (66%) patients were prescribed at least one PIM at discharge; 1,176 (49%) continued a PIM from prior to admission, and 755 (31%) were prescribed at least one new PIM. In the 30 days after discharge, 218 (9%) experienced an adverse drug event (ADE) and 862 (36%) visited the emergency department (ED), were rehospitalized, or died. After adjustment, each additional new PIM and continued community PIM were respectively associated with a 21% (odds ratio [OR] = 1.21; 95% confidence interval [CI] = 1.01‐1.45) and a 10% (OR = 1.10; 95% CI = 1.01‐1.21) increased odds of ADEs. They were also respectively associated with a 13% (hazard ratio [HR] = 1.13; 95% CI = 1.03‐1.26) and a 5% (HR = 1.05; 95% CI = 1.00‐1.10) increased risk of ED visits, rehospitalization, and death. Two in three hospitalized patients were prescribed a PIM at discharge, and increasing numbers of PIMs were associated with an increased risk of ADEs and all‐cause adverse events. Improving hospital prescribing practices may reduce the frequency of PIMs and associated adverse events. J Am Geriatr Soc 68:1184–1192, 2020.
2020-03-30
Journal of the American Geriatrics Society (published)
doi.org
Towards Lifelong Self-Supervision For Unpaired Image-to-Image Translation
Victor Schmidt
Makesh Narsimhan Sreedhar
Mostafa Elaraby
Irina Rish
Unpaired Image-to-Image Translation (I2IT) tasks often suffer from lack of data, a problem which self-supervised learning (SSL) has recently… (see more) been very popular and successful at tackling. Leveraging auxiliary tasks such as rotation prediction or generative colorization, SSL can produce better and more robust representations in a low data regime. Training such tasks along an I2IT task is however computationally intractable as model size and the number of task grow. On the other hand, learning sequentially could incur catastrophic forgetting of previously learned tasks. To alleviate this, we introduce Lifelong Self-Supervision (LiSS) as a way to pre-train an I2IT model (e.g., CycleGAN) on a set of self-supervised auxiliary tasks. By keeping an exponential moving average of past encoders and distilling the accumulated knowledge, we are able to maintain the network's validation performance on a number of tasks without any form of replay, parameter isolation or retraining techniques typically used in continual learning. We show that models trained with LiSS perform better on past tasks, while also being more robust than the CycleGAN baseline to color bias and entity entanglement (when two entities are very close).
2020-03-30
ArXiv (preprint)
arxiv.org
arxiv.org
Atypical brain asymmetry in autism – a candidate for clinically meaningful stratification
Dorothea L. Floris
Thomas Wolfers
Mariam Zabihi
Nathalie E. Holz
Christine Ecker
Flavio Dell’Acqua
Simon Baron-Cohen
Rosemary Holt
Sarah Durston
Eva Loth
Andre Marquand
Christian Beckmann
Jumana Ahmad
Sara Ambrosino
Bonnie Auyeung
Tobias Banaschewski
Sarah Baumeister
Sven Bölte
Thomas Bourgeron
Carsten Bours … (see 51 more)
Michael Brammer
Daniel Brandeis
Claudia Brogna
Yvette de Bruijn
Jan K. Buitelaar
Bhismadev Chakrabarti
Tony Charman
Ineke Cornelissen
Daisy Crawley
Guillaume Dumas
Jessica Faulkner
Vincent Frouin
Pilar Garcés
David Goyard
Lindsay Ham
Hannah Hayward
Joerg F. Hipp
Mark Johnson
Emily J. H. Jones
Prantik Kundu
Meng-Chuan Lai
Xavier Liogier D’ardhuy
Michael V. Lombardo
David J. Lythgoe
René Mandl
Luke Mason
Maarten Mennes
Andreas Meyer-Lindenberg
Carolin Moessnang
Nico Mueller
Declan Murphy
Beth Oakley
Larry O’Dwyer
Marianne Oldehinkel
Bob Oranje
Gahan Pandina
Antonio Persico
Barbara Ruggeri
Amber N. V. Ruigrok
Jessica Sabet
Roberto Sacco
Antonia San José Cáceres
Emily Simonoff
Will Spooren
Julian Tillmann
Roberto Toro
Heike Tost
Jack Waldman
Steve C. R. Williams
Caroline Wooldridge
Marcel P. Zwiers
2020-03-24
bioRxiv (preprint)
doi.org
Overview of the TREC 2019 Fair Ranking Track
Asia J. Biega
Fernando Diaz
Michael D. Ekstrand
Sebastian Kohlmeier
2020-03-24
ArXiv (preprint)
arxiv.org
arxiv.org
Pix2Shape – Towards Unsupervised Learning of 3D Scenes from Images using a View-based Representation
Sai Rajeswar
Fahim Mannan
Florian Golemo
Jérôme Parent-lévesque
David Vázquez
Derek Nowrouzezahrai
Aaron Courville
We infer and generate three-dimensional (3D) scene information from a single input image and without supervision. This problem is under-expl… (see more)ored, with most prior work relying on supervision from, e.g., 3D ground-truth, multiple images of a scene, image silhouettes or key-points. We propose Pix2Shape, an approach to solve this problem with four components: (i) an encoder that infers the latent 3D representation from an image, (ii) a decoder that generates an explicit 2.5D surfel-based reconstruction of a scene from the latent code (iii) a differentiable renderer that synthesizes a 2D image from the surfel representation, and (iv) a critic network trained to discriminate between images generated by the decoder-renderer and those from a training distribution. Pix2Shape can generate complex 3D scenes that scale with the view-dependent on-screen resolution, unlike representations that capture world-space resolution, i.e., voxels or meshes. We show that Pix2Shape learns a consistent scene representation in its encoded latent space and that the decoder can then be applied to this latent representation in order to synthesize the scene from a novel viewpoint. We evaluate Pix2Shape with experiments on the ShapeNet dataset as well as on a novel benchmark we developed, called 3D-IQTT, to evaluate models based on their ability to enable 3d spatial reasoning. Qualitative and quantitative evaluation demonstrate Pix2Shape's ability to solve scene reconstruction, generation, and understanding tasks.
2020-03-19
International Journal of Computer Vision (unknown)
doi.org
arxiv.org
Multinational Investigation of Fracture Risk with Antidepressant Use by Class, Drug, and Indication
Robyn Tamblyn
David W. Bates
David L. Buckeridge
William G. Dixon
Nadyne Girard
Jennifer S. Haas
Bettina Habib
Usman Iqbal
Jack Li
Therese Sheppard
Antidepressants increase the risk of falls and fracture in older adults. However, risk estimates vary considerably even in comparable popula… (see more)tions, limiting the usefulness of current evidence for clinical decision making. Our aim was to apply a common protocol to cohorts of older antidepressant users in multiple jurisdictions to estimate fracture risk associated with different antidepressant classes, drugs, doses, and potential treatment indications. Retrospective (2009–2014) cohort study. Five jurisdictions in the United States, Canada, United Kingdom, and Taiwan. Older antidepressant users—subjects were followed from first antidepressant prescription or dispensation to first fracture or until the end of follow‐up. The risk of fractures with antidepressants was estimated by multivariable Cox proportional hazards models using time‐varying measures of antidepressant dose and use vs nonuse, adjusting for patient characteristics. Between 42.9% and 55.6% of study cohorts were 75 years and older, and 29.3% to 45.4% were men. Selective serotonin reuptake inhibitors (SSRIs) (48.4%‐60.0%) were the predominant class used in North America compared with tricyclic antidepressants (TCAs) in the United Kingdom and Taiwan (49.6%‐53.6%). Fracture rates varied from 37.67 to 107.18 per 1,000. The SSRIs citalopram (hazard ratio [HR] = 1.23; 95% confidence interval [CI] = 1.11‐1.36 to HR = 1.43; 95% CI = 1.11‐1.84) and sertraline (HR = 1.36; 95% CI = 1.10‐1.68), the SNRI duloxetine (HR = 1.41; 95% CI = 1.06‐1.88), TCAs doxepin (HR = 1.36; 95% CI = 1.00‐1.86) and imipramine (HR = 1.16; 95% CI = 1.05‐1.28), and atypicals (HR = 1.34; 95% CI = 1.14‐1.58) increased fracture risk in some but not all jurisdictions. In the United States and the United Kingdom, fracture risk with all classes was higher when prescribed for depression than chronic pain, a trend that is likely explained by drug choice. The fracture risk for patients may be reduced by selecting paroxetine, an SSRI with lower risk than citalopram, the SNRI venlafaxine over duloxetine, and the TCA amitriptyline over imipramine or doxepin. There is uncertainty about the risk associated with the atypical antidepressants. J Am Geriatr Soc 68:1494‐1503, 2020.
2020-03-16
Journal of the American Geriatrics Society (published)
doi.org
Bayesian latent multi‐state modeling for nonequidistant longitudinal electronic health records
Yu Luo
David A. Stephens
Aman Verma
David L. Buckeridge
Large amounts of longitudinal health records are now available for dynamic monitoring of the underlying processes governing the observations… (see more). However, the health status progression across time is not typically observed directly: records are observed only when a subject interacts with the system, yielding irregular and often sparse observations. This suggests that the observed trajectories should be modeled via a latent continuous‐time process potentially as a function of time‐varying covariates. We develop a continuous‐time hidden Markov model to analyze longitudinal data accounting for irregular visits and different types of observations. By employing a specific missing data likelihood formulation, we can construct an efficient computational algorithm. We focus on Bayesian inference for the model: this is facilitated by an expectation‐maximization algorithm and Markov chain Monte Carlo methods. Simulation studies demonstrate that these approaches can be implemented efficiently for large data sets in a fully Bayesian setting. We apply this model to a real cohort where patients suffer from chronic obstructive pulmonary disease with the outcome being the number of drugs taken, using health care utilization indicators and patient characteristics as covariates.
2020-03-10
Biometrics (published)
doi.org