The Role of Schemas in Reinforcement Learning: Insights and Implications for Generalization

In cognitive psychology, schemas are considered to be “building blocks” of cognition, shaping how people view the world and interact with it. The goal of this paper is to propose a method for learning schemas in RL. We argue that, by representing tasks through schemas, agents can more effectively generalize from past experiences and adapt to new, unseen environments with minimal data.

with Prof.Doina Precup and Prof. Blake Richards

Beyond Multitask learning in Continual Learning

Continual Learning solutions often treat multitask learning as an upper-bound of what the learning process can achieve. This is a natural assumption, given that this objective directly addresses the catastrophic forgetting problem, which has been a central focus in early works. However, depending on the nature of the distributional shift in the data, the multi-task solution is not always optimal for the broader continual learning problem. We draw on principles from online learning to formalize the limitations of multitask objectives.

with Giulia Lanzillotta, Prof. Claire Vernade, and Razvan Pascanu

Testing Causal Hypotheses Through Hierarchical Reinforcement Learning

we propose hierarchical reinforcement learning (HRL) as a key ingredient to building agents that can systematically generate and test hypotheses that enables transferrable learning of the world, and discuss potential implementation strategies.

Presented at Intrinsically Motivated Open-ended Learning - IMOL Workshop at NeurIPS 2024 as a poster.

with Dongyan Lin and Anthony Chen

Towards Efficient Generalization in Continual RL using Episodic Memory

As an awardee of Microsoft Research Research grant, with Prof. Blake Richards, Prof. Guillaume Lajoie, Mehdi Fatemi, Ida Momennejad, Prof. Geoffrey J. Gordon, and John Langford. Gave an invited talk at Microsoft Research Summit 2021. Slides.

Possible Principles for Aligned Structure Learning Agents

This paper offers a roadmap for the development of scalable aligned artificial intelligence (AI) from first principle descriptions of natural intelligence. Read More.

with Lancelot Da Costa and Cristian Dragos-Manta

The Cancer Genome Atlas Program (TCGA) Clinical Benchmark

This paper provide a benchmark to study cancer and gene expression relations. Tasks are a combination of clinical features and cancer study. An example of a task would be predicting gender, age, alcohol document history, family history of stomach cancer, the code of the disease and other clinical attributes for different types of patients based on their gene expressions. Read More. Github.

with Joseph Paul Cohen and Prof. Thomas Fevens

Torchmeta: A meta-learning library for pytorch

We introduce Torchmeta, a library built on top of PyTorch that enables seamless and consistent evaluation of meta-learning algorithms on multiple datasets, by providing data-loaders for most of the standard benchmarks in few-shot classification and regression, with a new meta-dataset abstraction. Read More. Github.

with Tristan Deleu and Prof. Yoshua Bengio