- The introduction of stochasticity in the Arcade Learning Environment, the research interface to Atari 2600 games, as well as some methodological best practices to aid reproducibility. See for instance this paper.
- Eigenoptions, which are options discovered through time-based representations and that have been shown to improve exploration in reinforcement learning. See for instance this and this paper.
- The use of time-based representations to quantify uncertainty, for example to drive an agent's exploratory behaviour. See for instance this paper.
I am a research scientist at Google Brain in Montréal, Canada. My research focuses on reinforcement learning, particularly on topics such as generalization, exploration, option discovery and reproducibility. I joined Google Brain after finishing my Ph.D. at the University of Alberta, where I worked with Michael Bowling and Marc G. Bellemare. During my Ph.D. I interned at Microsoft Research, IBM Research, and DeepMind. Some of my research contributions include: