Publications

Graphically Structured Diffusion Models

Christian Dietrich Weilbach

William Harvey

Frank N. Wood

2023-04-23

ICML.cc/2023/Conference (poster)

doi.org

proceedings.mlr.press

Hyena Hierarchy: Towards Larger Convolutional Language Models

Michael Poli

Stefano Massaroli

Eric Nguyen

Daniel Y Fu

Tri Dao

Stephen Baccus

Yoshua Bengio

Stefano Ermon

Christopher Re

2023-04-23

ICML.cc/2023/Conference (poster)

doi.org

proceedings.mlr.press

Interventional Causal Representation Learning

Kartik Ahuja

Yixin Wang

Divyat Mahajan

Yoshua Bengio

Causal representation learning seeks to extract high-level latent factors from low-level sensory data. Most existing methods rely on observa… (voir plus)tional data and structural assumptions (e.g., conditional independence) to identify the latent factors. However, interventional data is prevalent across applications. Can interventional data facilitate causal representation learning? We explore this question in this paper. The key observation is that interventional data often carries geometric signatures of the latent factors' support (i.e. what values each latent can possibly take). For example, when the latent factors are causally connected, interventions can break the dependency between the intervened latents' support and their ancestors'. Leveraging this fact, we prove that the latent causal factors can be identified up to permutation and scaling given data from perfect

2023-04-23

ICML.cc/2023/Conference (poster)

doi.org

proceedings.mlr.press

Target-based Surrogates for Stochastic Optimization

Jonathan Wilder Lavington

Sharan Vaswani

Reza Babanezhad Harikandeh

Mark Schmidt

Nicolas Roux

We consider minimizing functions for which it is expensive to compute the gradient. Such functions are prevalent in reinforcement learning, … (voir plus)imitation learning and bilevel optimization. Our target optimization framework uses the (expensive) gradient computation to construct surrogate functions in a \emph{target space} (e.g. the logits output by a linear model for classification) that can be minimized efficiently. This allows for multiple parameter updates to the model, amortizing the cost of gradient computation. In the full-batch setting, we prove that our surrogate is a global upper-bound on the loss, and can be (locally) minimized using a black-box optimization algorithm. We prove that the resulting majorization-minimization algorithm ensures convergence to a stationary point of the loss. Next, we instantiate our framework in the stochastic setting and propose the

2023-04-23

ICML.cc/2023/Conference (poster)

doi.org

proceedings.mlr.press

Environmental Scan of Existing Digital Health Solutions for Older Adults Living with Neurocognitive Disorders (Mild and Major) and Their Informal Caregivers: Summary Report.

Ambily Jose

Maxime Sasseville

Ellen Gorus

Anik Giguère

Anne Bourbonnais

Samira Abbasgholizadeh Rahimi

Ronald Buyl

Marie-Pierre Gagnon

: Digital health has added numerous promising solutions to enhance the health and wellness of people living with dementia and other cognitiv… (voir plus)e problems and their informal caregivers. This work aims to summarize currently available digital health solutions and their related characteristics to develop a decision support tool for older adults living with mild or major neurocognitive disorders and their informal caregivers. We conducted an environmental scan to identify digital health solutions from a systematic review and targeted searches for grey literature covering the regions of Canada and Europe. Technological tools were scanned based on a preformatted extraction grid. We assessed their relevance based on selected attributes. We identified 100 available digital health solutions. The majority (56%) were not specific to dementia. Only 28% provided scientific evidence of their effectiveness. Remote patient care, movement tracking and cognitive exercises were the most common purposes of digital health solutions. Most solutions were presented as mobility aid tools, pill dispensers, apps, web, or a combination of these platforms. This knowledge will inform the development of a decision support tool to assist older adults and their informal caregivers in their search for adequate eHealth solutions according to their needs and preferences, based on trustable information.

2023-04-21

Proceedings of the 9th International Conference on Information and Communication Technologies for Ageing Well and e-Health (publié)

doi.org

An exploratory cross-sectional study of the effects of ongoing relationships with accompanying patients on cancer care experience, self-efficacy, and psychological distress

Marie-Pascale Pomey

Monica Iliescu Nelea

Louise Normandin

Cécile Vialaron

Karine Bouchard

Marie-Andrée Côté

Maria Alejandra Rodriguez Duarte

Djahanchah Philip Ghadiri

Israël Fortin

Danielle Charpentier

Mélanie Lavoie-Tremblay

Nicolas Fernandez

Antoine Boivin

Michel Dorval

Mado Desforges

Catherine Régis

Isabelle Ganache

Lynda Bélanger

Zeev Rosberger

Michel Alain Danino … (voir 3 de plus)

Jean-François Pelletier

Thi Trinh Thuc Vu

Michèle de Guise

Centre hospitalier de l’Université de Montréal in Canada introduced accompanying patients (APs) into the breast cancer care trajectory. … (voir plus)APs are patients who have been treated for breast cancer and have been integrated into the clinical team to expand the services offered to people affected by cancer. This study describes the profiles of the people who received the support and explores whether one-offs vs ongoing encounters with APs influence their experience of care, on self-efficacy in coping with cancer, and on their level of psychological distress. An exploratory cross-sectional study was carried out among patients to compare patients who had one encounter with an AP (G1) with those who had had several encounters (G2). Five questionnaires were administered on socio-demographic characteristics, care pathway, evaluation of the support experience, self-efficacy in coping with cancer, and level of psychological distress. Logbooks, completed by the APs, determined the number of encounters. Linear regression models were used to evaluate the associations between the number of encounters, patient characteristics, care pathway, number of topics discussed, self-efficacy measures in coping with cancer, and level of psychological distress. Between April 2020 and December 2021, 60% of 535 patients who were offered support from an AP accepted. Of these, one hundred and twenty-four patients participated in the study. The study aimed to recruit a minimum of 70 patients with the expectation of obtaining at least 50 participants, assuming a response rate of 70%. There were no differences between G1 and G2 in terms of sociodemographic data and care pathways. Statistical differences were found between G1 and G2 for impacts on and the return to daily life (p = 0.000), the return to the work and impacts on professional life (p = 0.044), announcement of a diagnosis to family and friends (p = 0.033), and strategies for living with treatment under the best conditions (p = 0.000). Significant differences were found on the topics of cancer (p = 0.000), genetic testing (p = 0.023), therapeutic options (p = 0.000), fatigue following treatment (p = 0.005), pain and discomfort after treatment or surgery (p = 0.000), potential emotions and their management (p = 0.000) and the decision-making processes (p = 0.011). A significant relationship was found between the two groups for patients’ ability to cope with cancer (p = 0.038), and their level of psychological distress at different stages of the care pathway (p = 0.024). This study shows differences between one-time and ongoing support for cancer patients. It highlights the potential for APs to help patients develop self-efficacy and cope with the challenges of cancer treatment.

2023-04-21

BMC Cancer (publié)

doi.org

SSS3D: Fast Neural Architecture Search For Efficient Three-Dimensional Semantic Segmentation

Olivier Therrien

Marihan Amein

Zhuoran Xiong

Warren J. Gross

Brett Meyer

We present SSS3D, a fast multi-objective NAS framework designed to find computationally efficient 3D semantic scene segmentation networks. I… (voir plus)t uses RandLA-Net, an off-the-shelf point-based network, as a super-network to enable weight sharing and reduce search time by 99.67% for single-stage searches. SSS3D has a complex search space composed of sampling and architectural parameters that can form 2.88 * 10^17 possible networks. To further reduce search time, SSS3D splits the complete search space and introduces a two-stage search that finds optimal subnetworks in 54% of the time required by single-stage searches.

2023-04-20

ArXiv (prépublication)

doi.org

arxiv.org

The Flag and the Cross: White Christian Nationalism and the Threat to American Democracy by Philip S. Gorski and Samuel L. Perry (review)

David M. Krueger

2023-04-20

(publié)

www.semanticscholar.org

Aspirations and Practice of ML Model Documentation: Moving the Needle with Nudging and Traceability

Avinash Bhat

Austin Coursey

Grace Hu

Sixian Li

Nadia Nahar

Shurui Zhou

Christian Kästner

Jin L.C. Guo

The documentation practice for machine-learned (ML) models often falls short of established practices for traditional software, which impede… (voir plus)s model accountability and inadvertently abets inappropriate or misuse of models. Recently, model cards, a proposal for model documentation, have attracted notable attention, but their impact on the actual practice is unclear. In this work, we systematically study the model documentation in the field and investigate how to encourage more responsible and accountable documentation practice. Our analysis of publicly available model cards reveals a substantial gap between the proposal and the practice. We then design a tool named DocML aiming to (1) nudge the data scientists to comply with the model cards proposal during the model development, especially the sections related to ethics, and (2) assess and manage the documentation quality. A lab study reveals the benefit of our tool towards long-term documentation quality and accountability.

2023-04-18

Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (publié)

doi.org

arxiv.org

Co-Writing Screenplays and Theatre Scripts with Language Models: Evaluation by Industry Professionals

Piotr Mirowski

Kory Mathewson

Jaylen Pittman

Richard Evans

2023-04-18

Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (publié)

doi.org