Tom Van de Wiele
Papers
1
Total Citations
155
H-Index
1
About
Tom Van de Wiele is a leading researcher in reinforcement learning (RL), best known for pioneering methods that enable agents to master complex tasks from scratch under sparse reward conditions. His most influential work, "Learning by Playing - Solving Sparse Reward Tasks from Scratch" (2018, 155 citations), introduced Scheduled Auxiliary Control (SAC-X)—a paradigm that equips RL agents with a suite of general auxiliary tasks to bootstrap learning when primary rewards are rare or absent. This breakthrough has significantly advanced the field of hierarchical and intrinsically motivated RL, offering a practical path to solving long-horizon problems that previously stymied traditional algorithms. Van de Wiele’s contributions are central to the development of more autonomous and sample-efficient agents, with his work frequently cited in subsequent research on exploration, curriculum learning, and multi-task RL. His research continues to shape how machines learn from minimal feedback, making him a key figure in modern reinforcement learning.
Research Focus
Key Achievements
Top Papers
- 1Learning by Playing - Solving Sparse Reward Tasks from Scratch155 citations · 2018