Thomas Anthony
Papers
1
Total Citations
139
H-Index
1
About
Thomas Anthony is a leading researcher in artificial intelligence, with a primary focus on reinforcement learning, sequential decision-making, and the integration of deep learning with classical planning methods. His most influential contribution is the development of Expert Iteration (ExIt), a novel reinforcement learning algorithm introduced in his highly cited 2017 paper "Thinking Fast and Slow with Deep Learning and Tree Search" (139 citations). This work elegantly combines the intuitive, pattern-matching capabilities of deep neural networks with the deliberate, look-ahead reasoning of tree search, drawing inspiration from cognitive science's dual-process theory. ExIt has proven remarkably effective across diverse domains, including structured prediction, robotic control, and game playing, offering a principled framework for training policies that can both plan and generalise. Anthony's research has significantly advanced the field by demonstrating how to efficiently bridge the gap between fast, reactive policies and slow, deliberative planning, enabling more capable and sample-efficient AI systems. His work continues to influence modern reinforcement learning and remains essential reading for researchers tackling complex sequential decision-making problems.
Research Focus
Key Achievements
Top Papers
- 1Thinking Fast and Slow with Deep Learning and Tree Search139 citations · 2017