Raymond Chua

📍 Montréal, Québec, Canada

Last updated 20 Oct 2025.

I am a final-year PhD candidate at McGill University and Mila, supervised by Doina Precup and Blake Richards. My research bridges machine learning and computational neuroscience, exploring how neural mechanisms of representation and memory can inspire more adaptive and continual learning in artificial agents. Building on this foundation, I am increasingly interested in mechanistic interpretability, using techniques inspired by neuroscience — such as representation similarity analysis and cross-attention probing — to understand how predictive representations such as successor features encode and transform information over time. I’m expected to graduate in Fall 2025 and am seeking opportunities to study and design learning systems that bridge theory (reinforcement learning), scalable models (foundation and language models), and real-world decision-making (robotics, embodied AI, and autonomous systems).

Beyond research, I’m passionate about improving AI capabilities through academic–industry partnerships, where I mentor students from McGill, UdeM, and Mila as they tackle real-world challenges with companies seeking to integrate machine learning into their products and pipelines. During my free time, I enjoy pushing both my intellectual and physical abilities through triathlon, which continues to teach me about endurance, balance, and growth.

news

Nov 13, 2025	Visited Prof. Rui Ponte Costa’s lab at University of Oxford, where I gave a talk on my work titled “Brain-Inspired Continual Reinforcement Learning Agents.” Rui was my master thesis supervisor, and it was wonderful to reconnect with him and his lab again!
Nov 12, 2025	Met up with the team at Prima Mente which is neuroscience-AI startup in London, UK, focusing on understanding the brain and advancing potential treatments for neurological and neurodegenerative diseases.
Oct 20, 2025	Gave a talk titled “Brain-Inspired Reinforcement Learning Agents” to visiting researchers from the Lamarr Institute for Machine Learning and Artificial Intelligence (Germany) at Mila – Quebec AI Institute, Montréal, Canada.
Sep 28, 2025	Excited to share that my first collaborative work, Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments , with Riley Simmons-Edler, Ryan P. Badman, Felix Baastad Berg, John J. Vastola, Joshua Lunger, William Qian, and Kanaka Rajan, has been accepted to NeurIPS 2025! We developed a foraging environment (based on Craftax) and trained PPO agents with compact GRUs, uncovering that even model-free agents can plan and recall across hundreds of timesteps.
Mar 10, 2025	Invited by the Integrate and Fire Seminar organisers at McGill University to share my perspective on the discussion theme of the night, “How Intelligent is AI?” Through sharing my own work on reinforcement learning, I presented some evidence which showed that agents learned solely on rewards can obtain some form of intelligence.

selected publications

NeurIPS

Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments

Riley Simmons-Edler, Ryan P Badman, Felix Baastad Berg, and 5 more authors

Proceedings of the 39th Conference on Neural Information Processing Systems (NeurIPS), 2025

This is my first collaboration work with members of Prof. Kanaka Rajan’s lab. Riley and Ryan are first authors, and Prof. Kanaka Rajan is the corresponding author.

PDF
NeurIPS

Learning Successor Features the Simple Way

Raymond Chua, Arna Ghosh, Christos Kaplanis, and 2 more authors

Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS), 2024

Raymond Chua is the corresponding author. Blake A. Richards and Doina Precup are co-senior authors.

PDF Code
Journal

Learning offline: memory replay in biological and artificial reinforcement learning

Emma L. Roscow, Raymond Chua, Rui Ponte Costa, and 2 more authors

Trends in Neurosciences, 2021

Abs PDF

Learning to act in an environment to maximise rewards is among the brain’s key functions. This process has often been conceptualised within the framework of reinforcement learning, which has also gained prominence in machine learning and artificial intelligence (AI) as a way to optimise decision making. A common aspect of both biological and machine reinforcement learning is the reactivation of previously experienced episodes, referred to as replay. Replay is important for memory consolidation in biological neural networks and is key to stabilising learning in deep neural networks. Here, we review recent developments concerning the functional roles of replay in the fields of neuroscience and AI. Complementary progress suggests how replay might support learning processes, including generalisation and continual learning, affording opportunities to transfer knowledge across the two fields to advance the understanding of biological and artificial learning and memory.