Krishnamurthy Dvijotham
Papers
2
Total Citations
394
H-Index
2
About
Krishnamurthy Dvijotham is a leading researcher in reinforcement learning (RL) and control theory, with a focus on safe and efficient decision-making for physical systems. His work bridges the gap between theoretical guarantees and real-world deployment, particularly in high-stakes environments like data center cooling and robotics. In his highly cited 2018 paper, "Safe Exploration in Continuous Action Spaces" (275 citations), Dvijotham introduced a method for RL agents to never violate critical constraints during exploration, leveraging the smooth dynamics of physical systems—a breakthrough for deploying AI in safety-critical applications. Earlier, his 2010 work on "Inverse Optimal Control with Linearly-Solvable MDPs" (119 citations) advanced inverse reinforcement learning by recovering not just the expert’s policy, but also the underlying value and cost functions, offering a more complete understanding of behavior. Dvijotham’s contributions have been instrumental in making RL practical and trustworthy, earning him recognition as a key figure in safe AI. His research continues to shape how autonomous systems learn and operate under constraints, with broad implications for robotics, energy management, and beyond.
Research Focus
Key Achievements
Top Papers
- 1Safe Exploration in Continuous Action Spaces275 citations · 2018
- 2Inverse Optimal Control with Linearly-Solvable MDPs119 citations · 2010