Publications

Performance-gated deliberation: A context-adapted strategy in which urgency is opportunity cost
Maximilian Puelma Touzel
Paul Cisek
Guillaume Lajoie
Finding the right amount of deliberation, between insufficient and excessive, is a hard decision making problem that depends on the value we… (voir plus) place on our time. Average-reward, putatively encoded by tonic dopamine, serves in existing reinforcement learning theory as the opportunity cost of time, including deliberation time. Importantly, this cost can itself vary with the environmental context and is not trivial to estimate. Here, we propose how the opportunity cost of deliberation can be estimated adaptively on multiple timescales to account for non-stationary contextual factors. We use it in a simple decision-making heuristic based on average-reward reinforcement learning (AR-RL) that we call Performance-Gated Deliberation (PGD). We propose PGD as a strategy used by animals wherein deliberation cost is implemented directly as urgency, a previously characterized neural signal effectively controlling the speed of the decision-making process. We show PGD outperforms AR-RL solutions in explaining behaviour and urgency of non-human primates in a context-varying random walk prediction task and is consistent with relative performance and urgency in a context-varying random dot motion task. We make readily testable predictions for both neural activity and behaviour.
2022-04-30
PLoS Comput. Biol. (publié)
doi.org
Predicting the probability distribution of bus travel time to measure the reliability of public transport services
Léa Ricard
Guy Desaulniers
Andrea Lodi
Louis-Martin Rousseau
2022-04-30
Transportation Research Part C: Emerging Technologies (publié)
doi.org
Subgraph Retrieval Enhanced Model for Multi-hop Knowledge Base Question Answering
Jing Zhang
Xiaokang Zhang
Jifan Yu
Jian Tang
Jie Tang
Cuiping Li
Hong Chen
Recent works on knowledge base question answering (KBQA) retrieve subgraphs for easier reasoning. A desired subgraph is crucial as a small o… (voir plus)ne may exclude the answer but a large one might introduce more noises. However, the existing retrieval is either heuristic or interwoven with the reasoning, causing reasoning on the partial subgraphs, which increases the reasoning bias when the intermediate supervision is missing. This paper proposes a trainable subgraph retriever (SR) decoupled from the subsequent reasoning process, which enables a plug-and-play framework to enhance any subgraph-oriented KBQA model. Extensive experiments demonstrate SR achieves significantly better retrieval and QA performance than existing retrieval methods. Via weakly supervised pre-training as well as the end-to-end fine-tuning, SRl achieves new state-of-the-art performance when combined with NSM, a subgraph-oriented reasoner, for embedding-based KBQA methods.
2022-04-30
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (publié)
doi.org
arxiv.org
On the estimation of discrete choice models to capture irrational customer behaviors
Sanjay Dominik Jena
Andrea Lodi
Claudio Sole
The random utility maximization model is by far the most adopted framework to estimate consumer choice behavior. However, behavioral economi… (voir plus)cs has provided strong empirical evidence of irrational choice behaviors, such as halo effects, that are incompatible with this framework. Models belonging to the random utility maximization family may therefore not accurately capture such irrational behavior. Hence, more general choice models, overcoming such limitations, have been proposed. However, the flexibility of such models comes at the price of increased risk of overfitting. As such, estimating such models remains a challenge. In this work, we propose an estimation method for the recently proposed generalized stochastic preference choice model, which subsumes the family of random utility maximization models and is capable of capturing halo effects. In particular, we propose a column-generation method to gradually refine the discrete choice model based on partially ranked preference sequences. Extensive computational experiments indicate that our model, explicitly accounting for irrational preferences, can significantly boost the predictive accuracy on both synthetic and real-world data instances. Summary of Contribution: In this work, we propose an estimation method for the recently proposed generalized stochastic preference choice model, which subsumes the family of random utility maximization models and is capable of capturing halo effects. Specifically, we show how to use partially ranked preferences to efficiently model rational and irrational customer types from transaction data. Our estimation procedure is based on column generation, where relevant customer types are efficiently extracted by expanding a treelike data structure containing the customer behaviors. Furthermore, we propose a new dominance rule among customer types whose effect is to prioritize low orders of interactions among products. An extensive set of experiments assesses the predictive accuracy of the proposed approach by comparing it against rank-based methods with only rational preferences and with more general benchmarks from the literature. Our results show that accounting for irrational preferences can boost predictive accuracy by 12.5% on average when tested on a real-world data set from a large chain of grocery and drug stores.
2022-04-30
INFORMS Journal on Computing (publié)
doi.org
arxiv.org
The Power of Prompt Tuning for Low-Resource Semantic Parsing
Nathan Schucher
Siva Reddy
Harm de Vries
Prompt tuning has recently emerged as an effective method for adapting pre-trained language models to a number of language understanding and… (voir plus) generation tasks. In this paper, we investigate prompt tuning for semantic parsing—the task of mapping natural language utterances onto formal meaning representations. On the low-resource splits of Overnight and TOPv2, we find that a prompt tuned T5-xl significantly outperforms its fine-tuned counterpart, as well as strong GPT-3 and BART baselines. We also conduct ablation studies across different model scales and target representations, finding that, with increasing model scale, prompt tuned T5 models improve at generating target representations that are far from the pre-training distribution.
2022-04-30
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) (publié)
doi.org
arxiv.org
Unsupervised Dependency Graph Network
Yikang Shen
Shawn Tan
Alessandro Sordoni
Peng Li
Jie Zhou
Aaron Courville
2022-04-30
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (publié)
doi.org
Using Interactive Feedback to Improve the Accuracy and Explainability of Question Answering Systems Post-Deployment
Zichao Li
Prakhar Sharma
Xing Han Lu
Jackie CK Cheung
Siva Reddy
2022-04-30
Findings of the Association for Computational Linguistics: ACL 2022 (publié)
doi.org
arxiv.org
Using Population Datasets to Identify the Brain Basis of Social Isolation
Danilo Bzdok
2022-04-30
Biological Psychiatry (publié)
doi.org
Public Perspectives on Exposure Notification Apps: A Patient and Citizen Co-Designed Study
Esli Osmanlliu
Jesseca Paquette
Maria Alejandra Rodriguez Duarte
Sylvain Bédard
Nathalie de Marcellis-Warin
Majlinda Zhegu
Catherine Régis
Marie-Eve Bouthillier
Annie-Danielle Grenier
Paul Lewis
Marie-Pascale Pomey
Canada deployed a digital exposure notification app (COVID Alert) as a strategy to support manual contact tracing. Our aims are to (1) asses… (voir plus)s the use, knowledge, and concerns of the COVID Alert app, (2) identify predictors of app downloads, and (3) develop strategies to promote social acceptability. A 36-item questionnaire was co-designed by 12 citizens and patients partnered with 16 academic researchers and was distributed in the province of Québec, Canada, from May 27 to 28 June 2021. Of 959 respondents, 43% had downloaded the app. Messaging from government sources constituted the largest influence on app download. Infrequent social contacts and perceived app inefficacy were the main reasons not to download the app. Cybersecurity, data confidentiality, loss of privacy, and geolocation were the most frequent concerns. Nearly half of the respondents inaccurately believed that the app used geolocation. Most respondents supported citizen involvement in app development. The identified predictors for app uptake included nine characteristics. In conclusion, this project highlights four key themes on how to promote the social acceptability of such tools: (1) improved communication and explanation of key app characteristics, (2) design features that incentivize adoption, (3) inclusive socio-technical features, and (4) upstream public partnership in development and deployment.
2022-04-29
Journal of Personalized Medicine (publié)
doi.org
Determinants of technology adoption and continued use among cognitively impaired older adults: a qualitative study
Samantha Dequanter
Maaike Fobelets
Iris Steenhout
Marie-Pierre Gagnon
Anne Bourbonnais
S. A. Rahimi
Ronald Buyl
Ellen Gorus
2022-04-27
BMC Geriatrics (publié)
doi.org
Radiomics-Based Machine Learning for Outcome Prediction in a Multicenter Phase II Study of Programmed Death-Ligand 1 Inhibition Immunotherapy for Glioblastoma
Elizabeth George
Elizabeth Flagg
Kuan-chun Chang
Hai-Yang Bai
H. Aerts
Martin Vallières
David A. Reardon
R.Y. Huang
BACKGROUND AND PURPOSE: Imaging assessment of an immunotherapy response in glioblastoma is challenging due to overlap in the appearance of t… (voir plus)reatment-related changes with tumor progression. Our purpose was to determine whether MR imaging radiomics-based machine learning can predict progression-free survival and overall survival in patients with glioblastoma on programmed death-ligand 1 inhibition immunotherapy. MATERIALS AND METHODS: Post hoc analysis was performed of a multicenter trial on the efficacy of durvalumab in glioblastoma (n = 113). Radiomics tumor features on pretreatment and first on-treatment time point MR imaging were extracted. The random survival forest algorithm was applied to clinical and radiomics features from pretreatment and first on-treatment MR imaging from a subset of trial sites (n = 60–74) to train a model to predict long overall survival and progression-free survival and was tested externally on data from the remaining sites (n = 29–43). Model performance was assessed using the concordance index and dynamic area under the curve from different time points. RESULTS: The mean age was 55.2 (SD, 11.5) years, and 69% of patients were male. Pretreatment MR imaging features had a poor predictive value for overall survival and progression-free survival (concordance index  = 0.472–0.524). First on-treatment MR imaging features had high predictive value for overall survival (concordance index = 0.692–0.750) and progression-free survival (concordance index = 0.680–0.715). CONCLUSIONS: A radiomics-based machine learning model from first on-treatment MR imaging predicts survival in patients with glioblastoma on programmed death-ligand 1 inhibition immunotherapy.
2022-04-27
American Journal of Neuroradiology (publié)
doi.org
Current State and Future Directions for Learning in Biological Recurrent Neural Networks: A Perspective Piece
Luke Y. Prince
Roy Henha Eyono
Ellen Boven
Arna Ghosh
Joe Pemberton
Franz Scherr
Claudia Clopath
Rui Ponte Costa
Wolfgang Maass
Blake A. Richards
Cristina Savin
Katharina Anna Wilmes
We provide a brief review of the common assumptions about biological learning with findings from experimental neuroscience and contrast them… (voir plus) with the efficiency of gradient-based learning in recurrent neural networks. The key issues discussed in this review include: synaptic plasticity, neural circuits, theory-experiment divide, and objective functions. We conclude with recommendations for both theoretical and experimental neuroscientists when designing new studies that could help bring clarity to these issues.
2022-04-26
Neurons, Behavior, Data Analysis, and Theory (publié)
doi.org
arxiv.org