Publications

Optimistic critics can empower small actors
Olya Mastikhina
Dhruv Sreenivas
Pablo Samuel Castro
Actor-critic methods have been central to many of the recent advances in deep reinforcement learning. The most common approach is to use _sy… (see more)mmetric_ architectures, whereby both actor and critic have the same network topology and number of parameters. However, recent works have argued for the advantages of _asymmetric_ setups, specifically with the use of smaller actors. We perform broad empirical investigations and analyses to better understand the implications of this and find that, in general, smaller actors result in performance degradation and overfit critics. Our analyses suggest _poor data collection_, due to value underestimation, as one of the main causes for this behavior, and further highlight the crucial role the critic can play in alleviating this pathology. We explore techniques to mitigate the observed value underestimation, which enables further research in asymmetric actor-critic methods.
2025-05-08
rl-conference.cc/RLC/2025/Conference (published)
openreview.net
openreview.net
Sampling from Energy-based Policies using Diffusion
Vineet Jain
Tara Akhound-Sadegh
Siamak Ravanbakhsh
2025-05-08
rl-conference.cc/RLC/2025/Conference (published)
doi.org
openreview.net
Understanding Behavioral Metric Learning: A Large-Scale Study on Distracting Reinforcement Learning Environments
Ziyan "Ray" Luo
Tianwei Ni
Pierre-Luc Bacon
Doina Precup
Xujie Si
A key approach to state abstraction is approximating behavioral metrics (notably, bisimulation metrics) in the observation space and embeddi… (see more)ng these learned distances in the representation space. While promising for robustness to task-irrelevant noise, as shown in prior work, accurately estimating these metrics remains challenging, requiring various design choices that create gaps between theory and practice. Prior evaluations focus mainly on final returns, leaving the quality of learned metrics and the source of performance gains unclear. To systematically assess how metric learning works in deep reinforcement learning (RL), we evaluate five recent approaches, unified conceptually as isometric embeddings with varying design choices. We benchmark them with baselines across 20 state-based and 14 pixel-based tasks, spanning 370 task configurations with diverse noise settings. Beyond final returns, we introduce the evaluation of a denoising factor to quantify the encoder's ability to filter distractions. To further isolate the effect of metric learning, we propose and evaluate an isolated metric estimation setting, in which the encoder is influenced solely by the metric loss. Finally, we release an open-source, modular codebase to improve reproducibility and support future research on metric learning in deep RL.
2025-05-08
rl-conference.cc/RLC/2025/Conference (accepted)
doi.org
openreview.net
Can a Bayesian Oracle Prevent Harm from an Agent?
Yoshua Bengio
Michael K. Cohen
Nikolay Malkin
Matt MacDermott
Damiano Fornasiere
Pietro Greiner
Younesse Kaddar
Is there a way to design powerful AI systems based on machine learning methods that would satisfy probabilistic safety guarantees? With the … (see more)long-term goal of obtaining a probabilistic guarantee that would apply in every context, we consider estimating a context-dependent bound on the probability of violating a given safety specification. Such a risk evaluation would need to be performed at run-time to provide a guardrail against dangerous actions of an AI. Noting that different plausible hypotheses about the world could produce very different outcomes, and because we do not know which one is right, we derive bounds on the safety violation probability predicted under the true but unknown hypothesis. Such bounds could be used to reject potentially dangerous actions. Our main results involve searching for cautious but plausible hypotheses, obtained by a maximization that involves Bayesian posteriors over hypotheses. We consider two forms of this result, in the iid case and in the non-iid case, and conclude with open problems towards turning such theoretical results into practical AI guardrails.
2025-05-06
Conference on Uncertainty in Artificial Intelligence (poster)
doi.org
proceedings.mlr.press
Mock Deep Testing: Toward Separate Development of Data and Models for Deep Learning
Ruchira Manke
Mohammad Wardat
Foutse Khomh
Hridesh Rajan
While deep learning (DL) has permeated, and become an integral component of many critical software systems, today software engineering resea… (see more)rch hasn't explored how to separately test data and models that are integral for DL approaches to work effectively. The main challenge in independently testing these components arises from the tight dependency between data and models. This research explores this gap, introducing our methodology of mock deep testing for unit testing of DL applications. To enable unit testing, we introduce a design paradigm that decomposes the workflow into distinct, manageable components, minimizes sequential dependencies, and modularizes key stages of the DL. For unit testing these components, we propose modeling their dependencies using mocks. This modular approach facilitates independent development and testing of the components, ensuring comprehensive quality assurance throughout the development process. We have developed KUnit, a framework for enabling mock deep testing for the Keras library. We empirically evaluated KUnit to determine the effectiveness of mocks. Our assessment of 50 DL programs obtained from Stack Overflow and GitHub shows that mocks effectively identified 10 issues in the data preparation stage and 53 issues in the model design stage. We also conducted a user study with 36 participants using KUnit to perceive the effectiveness of our approach. Participants using KUnit successfully resolved 25 issues in the data preparation stage and 38 issues in the model design stage. Our findings highlight that mock objects provide a lightweight emulation of the dependencies for unit testing, facilitating early bug detection. Lastly, to evaluate the usability of KUnit, we conducted a post-study survey. The results reveal that KUnit is helpful to DL application developers, enabling them to independently test each component effectively in different stages.
2025-05-05
2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE) (published)
doi.org
arxiv.org
Kernel-Level Event-Based Performance Anomaly Detection in Software Systems under Varying Load Conditions
Anthonia Njoku
Heng Li
Foutse Khomh
2025-05-04
Companion of the 16th ACM/SPEC International Conference on Performance Engineering (published)
doi.org
Position: Humanity Faces Existential Risk from Gradual Disempowerment
Jan Kulveit
Raymond Douglas
Nora Ammann
Deger Turan
David M. Krueger
David Duvenaud
2025-05-04
ICML.cc/2025/Position_Paper_Track (poster)
proceedings.mlr.press
proceedings.mlr.press
The Search for Squawk: Agile Modeling in Bioacoustics
Vincent Dumoulin
Otilia Stretcu
Jenny Hamer
Lauren Harrell
Rob Laber
Hugo Larochelle
Bart Van Merriënboer
Amanda K. Navine
Patrick Hart
Ben Williams
Timothy A. C. Lamont
Tries B. Rasak
Mars Coral Restoration Team
Sheryn Brodie
Brendan Doohan
Philip Eichinski
Paul Roe
Lin Schwarzkopf
Tom Denton
2025-05-04
ArXiv (preprint)
arxiv.org
arxiv.org
"It was 80% me, 20% AI": Seeking Authenticity in Co-Writing with Large Language Models
Angel Hsing-Chi Hwang
Q. Vera Liao
Su Lin Blodgett
A.R. Olteanu
Adam Trischler
2025-05-01
Proceedings of the ACM on Human-Computer Interaction (published)
doi.org
arxiv.org
Addressing Concept Mislabeling in Concept Bottleneck Models Through Preference Optimization
Emiliano Penaloza
Tianyue H. Zhang
Laurent Charlin
Mateo Espinosa Zarlenga
Concept Bottleneck Models (CBMs) propose to enhance the trustworthiness of AI systems by constraining their decisions on a set of human-unde… (see more)rstandable concepts. However, CBMs typically assume that datasets contain accurate concept labels-an assumption often violated in practice, which we show can significantly degrade performance (by 25% in some cases). To address this, we introduce the Concept Preference Optimization (CPO) objective, a new loss function based on Direct Preference Optimization, which effectively mitigates the negative impact of concept mislabeling on CBM performance. We provide an analysis of key properties of the CPO objective, showing it directly optimizes for the concept's posterior distribution, and contrast it against Binary Cross Entropy (BCE), demonstrating that CPO is inherently less sensitive to concept noise. We empirically confirm our analysis by finding that CPO consistently outperforms BCE on three real-world datasets, both with and without added label noise. We make our code available on Github.
2025-04-30
ICML.cc/2025/Conference (poster)
doi.org
proceedings.mlr.press
ALAS: Measuring Latent Speech-Text Alignment For Spoken Language Understanding In Multimodal LLMs
Pooneh Mousavi
Yingzhi Wang
Mirco Ravanaelli
Yusuf Cem Sübakan
2025-04-30
arXiv (published)
doi.org
arxiv.org
Building spatial world models from sparse transitional episodic memories
Zizhan He
Maxime Daigle
Pouya Bashivan
Many animals possess a remarkable capacity to rapidly construct flexible mental models of their environments. These world models are crucial… (see more) for ethologically relevant behaviors such as navigation, exploration, and planning. The ability to form episodic memories and make inferences based on these sparse experiences is believed to underpin the efficiency and adaptability of these models in the brain. Here, we ask: Can a neural network learn to construct a spatial model of its surroundings from sparse and disjoint episodic memories? We formulate the problem in a simulated world and propose a novel framework, the Episodic Spatial World Model (ESWM), as a potential answer. We show that ESWM is highly sample-efficient, requiring minimal observations to construct a robust representation of the environment. It is also inherently adaptive, allowing for rapid updates when the environment changes. In addition, we demonstrate that ESWM readily enables near-optimal strategies for exploring novel environments and navigating between arbitrary points, all without the need for additional training.
2025-04-30
arXiv (published)
doi.org
arxiv.org