Bhaskar Ramasubramanian
Papers
1
Total Citations
13
H-Index
1
About
Bhaskar Ramasubramanian is a researcher at the forefront of human-in-the-loop reinforcement learning, with a focus on making autonomous agents more adaptable and intuitive to train. His most-cited work, "FRESH: Interactive Reward Shaping in High-Dimensional State Spaces using Human Feedback" (2020, 13 citations), introduces a novel framework that allows human trainers to provide real-time, interactive feedback to shape an agent’s reward function, even in complex, high-dimensional environments. This contribution bridges the gap between human intuition and machine learning, enabling more efficient and safer policy learning in domains like robotics and game AI. By demonstrating that human players can outperform automated reward structures in certain settings, Ramasubramanian’s work highlights the practical value of integrating human expertise into reinforcement learning pipelines. His research is particularly impactful for students and engineers seeking to build collaborative AI systems that learn from natural, non-expert guidance. Through FRESH and related projects, he continues to advance the frontier of interactive machine learning, making autonomous systems not only smarter but also more responsive to human intent.
Research Focus
Key Achievements
Top Papers
- 1