Our research focuses on the development of methods for AI systems that learn over long-term deployments. Most AI methods assume a fixed set of tasks and a static instance distribution that is known at training time and evaluation time. Our work breaks these assumptions, allowing for our methods to adapt to new tasks while deployed in novel domains, unlocking new directions of work in robotics and medicine.
We are frequently looking for passionate new PhD students, Postdocs, and Master students to join the team! See our open positions for more details.
Two papers accepted to the First Conference on Lifelong Learning Agents (CoLLAs).January 2022
Our paper "Modular Lifelong Reinforcement Learning via Neural Composition" was accepted to ICML.May 2021
Our paper "Sharing Less is More: Lifelong Learning in Deep Networks with Selective Layer Transfer" was accepted to ICML.March 2021
Jorge Mendez was awarded third place prize of the Two Sigma Diversity PhD Fellowship.