David Abel

Portrait [CV] [Git] [Scholar] david_abel@brown.edu


I'm a Ph.D candidate in Computer Science at Brown University focusing on Artificial Intelligence, advised by Prof. Michael Littman.


My research investigates the role of representation in intelligence; I seek a unifying mathematical theory for models of the world that support effective exploration, planning, generalization, and causal inference. I typically work with the paradigm of Reinforcement Learning, drawing on tools from computational learning theory, probability, complex systems, and information theory.

I also care deeply about responsible applications of AI to scientific and societal challenges. Most recently, fellow grad student Emily Reif and I developed Machine Learning techniques to improve the efficiency of solar panels in line with the mission of Computational Sustainability.

Recent Publications

RL and Abstraction

Near Optimal Behavior via Approximate State Abstraction
ICML 2016

(paper, arXiv version, slides)

We investigate what approximate abstractions preserve near-optimality for Reinforcement Learning agents. Joint work with Ellis Hershkowitz and Prof. Littman.

Agent-Agnost HRL

Agent-Agnostic Human-In-The-Loop Reinforcement Learning
NIPS Workshop on the Future of Interactive Learning Machines 2016

(paper, arXiv version)

We advance a method for incorporating a teacher's advice into an arbitrary RL agent's learning process, capturing prior approaches like reward shaping and action pruning. Joint work with Owain Evans, John Salvatier, and Andreas Stuhlm├╝ller.

Full list of papers here.


For fun, I'm a big fan of basketball, hiking, travel, fitness, snowboarding, and music (especially folk, trance, and progressive metal).

I'm an advocate of a few specific causes: sustainability efforts, existential risk minimization, space exploration, and improving the diversity, quality, and accessibility of STEM education.

Always up for a chat - shoot me an email if you'd like to discuss anything!