Publications

Trait‐matching models predict pairwise interactions across regions, not food web properties

Dominique Caron

Ulrich Brose

Miguel Lurgi

F. Guillaume Blanchet

Dominique Gravel

Food webs are essential for understanding how ecosystems function, but empirical data on the interactions that make up these ecological netw… (see more)orks are lacking for most taxa in most ecosystems. Trait‐based models can fill these data gaps, but their ability to do so has not been widely tested. We test how well these models can extrapolate to new ecological communities both in terms of pairwise predator–prey interactions and higher level food web attributes (i.e. species position, food web‐level properties).Canada, Europe, Tanzania.Current.Terrestrial vertebrates.We train trait‐based models of pairwise trophic interactions on four independent vertebrate food webs (Canadian tundra, Serengeti, alpine south‐eastern Pyrenees and Europe) and evaluate how well these models predict pairwise interactions and network properties of each food web.We find that, overall, trait‐based models predict most interactions and their absence correctly. Performance was best for training and testing on the same food web (AUC > 0.90) and declined with environmental and phylogenetic distances with the strongest loss of performance for the tundra‐Serengeti ecosystems (AUC > 0.75). Network metrics were less well‐predicted than single interactions by our models with predicted food webs being more connected, less modular, and with higher mean trophic levels than observed.Theory predicts that the variability observed in food webs can be explained by differences in trait distributions and trait‐matching relationships. Our finding that trait‐based models can predict many trophic interactions, even in contrasting environments, adds to the growing body of evidence that there are general constraints on interactions and that trait‐based methods can serve as a useful first approximation of food webs in unknown areas. However, food webs are more than the sum of their parts, and predicting network attributes will likely require models that simultaneously predict individual interactions and community constraints.

2024-01-29

Global Ecology and Biogeography (published)

doi.org

Trait‐matching models predict pairwise interactions across regions, not food web properties

Dominique Caron

Ulrich Brose

Miguel Lurgi

F. Guillaume Blanchet

Dominique Gravel

Laura J. Pollock

2024-01-29

Global Ecology and Biogeography (published)

doi.org

Efficient Data-Driven MPC for Demand Response of Commercial Buildings

Marie-Christine Par'e

Vasken Dermardiros

Antoine Lesage-Landry

Model predictive control (MPC) has been shown to significantly improve the energy efficiency of buildings while maintaining thermal comfort.… (see more) Data-driven approaches based on neural networks have been proposed to facilitate system modelling. However, such approaches are generally nonconvex and result in computationally intractable optimization problems. In this work, we design a readily implementable energy management method for small commercial buildings. We then leverage our approach to formulate a real-time demand bidding strategy. We propose a data-driven and mixed-integer convex MPC which is solved via derivative-free optimization given a limited computational time of 5 minutes to respect operational constraints. We consider rooftop unit heating, ventilation, and air conditioning systems with discrete controls to accurately model the operation of most commercial buildings. Our approach uses an input convex recurrent neural network to model the thermal dynamics. We apply our approach in several demand response (DR) settings, including a demand bidding, a time-of-use, and a critical peak rebate program. Controller performance is evaluated on a state-of-the-art building simulation. The proposed approach improves thermal comfort while reducing energy consumption and cost through DR participation, when compared to other data-driven approaches or a set-point controller.

2024-01-28

ArXiv (preprint)

doi.org

arxiv.org

Graphylo: A deep learning approach for predicting regulatory DNA and RNA sites from whole-genome multiple alignments

Dongjoon Lim

Changhyun Baek

Mathieu Blanchette

2024-01-26

iScience (published)

doi.org

Black-Box Access is Insufficient for Rigorous AI Audits

Stephen Casper

Carson Ezell

Charlotte Siegmann

Noam Kolt

Taylor Lynn Curtis

Benjamin Bucknall

Andreas Haupt

Kevin Wei

Jérémy Scheurer

Marius Hobbhahn

Lee Sharkey

Satyapriya Krishna

Marvin Von Hagen

Silas Alberti

Alan Chan

Qinyi Sun

Michael Gerovitch

David Bau

Max Tegmark

David Scott Krueger … (see 1 more)

Dylan Hadfield-Menell

External audits of AI systems are increasingly recognized as a key mechanism for AI governance. The effectiveness of an audit, however, depe… (see more)nds on the degree of access granted to auditors. Recent audits of state-of-the-art AI systems have primarily relied on black-box access, in which auditors can only query the system and observe its outputs. However, white-box access to the system’s inner workings (e.g., weights, activations, gradients) allows an auditor to perform stronger attacks, more thoroughly interpret models, and conduct fine-tuning. Meanwhile, outside-the-box access to training and deployment information (e.g., methodology, code, documentation, data, deployment details, findings from internal evaluations) allows auditors to scrutinize the development process and design more targeted evaluations. In this paper, we examine the limitations of black-box audits and the advantages of white- and outside-the-box audits. We also discuss technical, physical, and legal safeguards for performing these audits with minimal security risks. Given that different forms of access can lead to very different levels of evaluation, we conclude that (1) transparency regarding the access and methods used by auditors is necessary to properly interpret audit results, and (2) white- and outside-the-box access allow for substantially more scrutiny than black-box access alone.

2024-01-25

ArXiv (preprint)

doi.org

arxiv.org

Black-Box Access is Insufficient for Rigorous AI Audits

Stephen Casper

Carson Ezell

Charlotte Siegmann

Noam Kolt

Taylor Lynn Curtis

Benjamin Bucknall

Andreas Haupt

Kevin Wei

Jérémy Scheurer

Marius Hobbhahn

Lee Sharkey

Satyapriya Krishna

Marvin Von Hagen

Silas Alberti

Alan Chan

Qinyi Sun

Michael Gerovitch

David Bau

Max Tegmark

David Scott Krueger … (see 1 more)

Dylan Hadfield-Menell

External audits of AI systems are increasingly recognized as a key mechanism for AI governance. The effectiveness of an audit, however, depe… (see more)nds on the degree of access granted to auditors. Recent audits of state-of-the-art AI systems have primarily relied on black-box access, in which auditors can only query the system and observe its outputs. However, white-box access to the system’s inner workings (e.g., weights, activations, gradients) allows an auditor to perform stronger attacks, more thoroughly interpret models, and conduct fine-tuning. Meanwhile, outside-the-box access to training and deployment information (e.g., methodology, code, documentation, data, deployment details, findings from internal evaluations) allows auditors to scrutinize the development process and design more targeted evaluations. In this paper, we examine the limitations of black-box audits and the advantages of white- and outside-the-box audits. We also discuss technical, physical, and legal safeguards for performing these audits with minimal security risks. Given that different forms of access can lead to very different levels of evaluation, we conclude that (1) transparency regarding the access and methods used by auditors is necessary to properly interpret audit results, and (2) white- and outside-the-box access allow for substantially more scrutiny than black-box access alone.

2024-01-25

ArXiv (preprint)

doi.org

arxiv.org

Black-Box Access is Insufficient for Rigorous AI Audits

Stephen Casper

Carson Ezell

Charlotte Siegmann

Noam Kolt

Taylor Lynn Curtis

Benjamin Bucknall

Andreas Haupt

Kevin Wei

Jérémy Scheurer

Marius Hobbhahn

Lee Sharkey

Satyapriya Krishna

Marvin Von Hagen

Silas Alberti

Alan Chan

Qinyi Sun

Michael Gerovitch

David Bau

Max Tegmark

David Scott Krueger … (see 1 more)

Dylan Hadfield-Menell

External audits of AI systems are increasingly recognized as a key mechanism for AI governance. The effectiveness of an audit, however, depe… (see more)nds on the degree of access granted to auditors. Recent audits of state-of-the-art AI systems have primarily relied on black-box access, in which auditors can only query the system and observe its outputs. However, white-box access to the system’s inner workings (e.g., weights, activations, gradients) allows an auditor to perform stronger attacks, more thoroughly interpret models, and conduct fine-tuning. Meanwhile, outside-the-box access to training and deployment information (e.g., methodology, code, documentation, data, deployment details, findings from internal evaluations) allows auditors to scrutinize the development process and design more targeted evaluations. In this paper, we examine the limitations of black-box audits and the advantages of white- and outside-the-box audits. We also discuss technical, physical, and legal safeguards for performing these audits with minimal security risks. Given that different forms of access can lead to very different levels of evaluation, we conclude that (1) transparency regarding the access and methods used by auditors is necessary to properly interpret audit results, and (2) white- and outside-the-box access allow for substantially more scrutiny than black-box access alone.

2024-01-25

ArXiv (preprint)

doi.org

arxiv.org

Human-Centered AI

Catherine Régis

Jean-Louis Denis

Maria Luciana Axente

Atsuo Kishimoto

2024-01-25

(published)

doi.org