Presented SHIRE at ICRA 2025!

Research Update! My paper titled “SHIRE: Enhancing Sample Efficiency using Human Intuition in REinforcement Learning”, co-authored with Adarsh Kumar Kosta and Kaushik Roy is going to be presented at IEEE International Conference on Robotics and Automation (ICRA) 2025!

This work addresses the well-known challenge of poor sample efficiency in reinforcement learning (RL) algorithms. Our approach leverages Probabilistic Graphical Models (PGMs) to incorporate task-specific human intuition directly into the RL training process, thereby significantly enhancing sample efficiency and highlighting the potential of human-guided RL for robotics.

If you want to learn more about this work, or just want to talk about Reinforcement Learning, Robotics, Robot Learning, RLHF in general, please join me in Atlanta, Georgia from 19-23 May!

---

For more information check out the paper here and the accompanying code