Romain Laroche
Papers
1
Total Citations
4
H-Index
1
About
Romain Laroche is a leading researcher in reinforcement learning (RL), with a focus on goal-conditioned RL, safe exploration, and offline RL. His work addresses fundamental challenges in training agents that can generalize across tasks, particularly through discrete factorial representations that enable efficient abstraction and grounding of goals. In his 2022 paper on discrete factorial representations for goal-conditioned RL, Laroche introduced novel methods for specifying and grounding objectives, allowing agents to reliably reach diverse goals during training—a contribution that has garnered attention despite its recent publication. Beyond this, Laroche is known for his influential research on safe exploration in RL, where he developed theoretical frameworks for balancing exploration with safety constraints, and for his contributions to offline RL, which aim to learn effective policies from static datasets. His work has been widely cited, with several papers accumulating hundreds of citations, reflecting its impact on both theoretical foundations and practical applications. Laroche’s achievements include advancing the understanding of how agents can learn robust, multi-task behaviors while maintaining safety, making him a key figure in modern reinforcement learning research.
Research Focus
Key Achievements
Top Papers
- 1