David Szepesvari

I am a PhD student at the Computing Science Department of the University of Alberta with a focus on Reinforcement Learning, supervised by Dale Schuurmans.

My research interests include continual learning and, temporal abstractions and model-based methods.

Before this, I worked at Google DeepMind as a Research Engineer, working across a number of domains with a focus on reinforcement learning. I completed my Masters in the Computational Statistics group at the University of Waterloo in 2015, under the supervision of Shai Ben-David. My thesis, "A Statistical Analysis of the Aggregation of Crowdsourced Labels", can be found here.

I enjoy mentoring and teaching, and have had the pleasure of doing so through the Eastern European Machine Learning School, AI4Good Lab, and UR2PhD, amongst others.

Outside of AI, I enjoy trying my hand at building physical things and racing my car at local RallyCross events. I also enjoy hiking, biking, and climbing.

Publications and Preprints

Mitigating the curse of horizon in Monte-Carlo returns.
A. Ayoub, D. Szepesvari, F. Zanini, B. Chan, D. Gupta, B.C. da Silva, D. Schuurmans, RLC, 2024
Reward-Respecting Subtasks for Model-Based Reinforcement Learning
R.S. Sutton, M.C. Machado, G.Z. Holland, F. Timbers, D. Szepesvari, B. Tanner, A. White, Artificial Intelligence, 2023, 324, 104001
When should agents explore?
M. Pislar, D. Szepesvari, G. Ostrovski, D. Borsa, T. Schaul, ICRL, 2022
Adapting behaviour for learning progress
T. Schaul, D. Borsa, D. Ding, David Szepesvari, G. Ostrovski, W. Dabney, S. Osindero, ArXiv, 2019
Grounded language learning in a simulated 3D world
K.M. Hermann, F. Hill, S. Green, F. Wang, R. Faulkner, H. Soyer, D. Szepesvari, W.M. Czarnecki, M. Jaderberg, D. Teplyashin, M. Wainwright, C. Apps, D. Hassabis, P. Blunsom, ArXiv, 2017
Attend, infer, repeat: Fast scene understanding with generative models
S.M. Eslami, N. Heess, T. Weber, Y. Tassa, D. Szepesvari, G.E. Hinton, NeurIPS, 2016

You may find my Google Scholar profile here.