Jens Erik Kveen

Norwegian University of Science and Technology

Papers

Total Citations

H-Index

About

Jens Erik Kveen is a researcher advancing the frontiers of model-based reinforcement learning (RL) for real-world robotic systems. His work directly tackles two critical bottlenecks in the field: sample efficiency and model-bias, which often prevent RL algorithms from being practically deployed. In his most-cited paper, "Addressing Sample Efficiency and Model-bias in Model-based Reinforcement Learning" (2022, 4 citations), Kveen proposes novel strategies to mitigate the compounding errors that arise when learned models deviate from reality, thereby improving the reliability and data economy of RL agents. This contribution is pivotal for transitioning RL from simulation to physical robots, where every interaction is costly. Kveen’s research is notable for its pragmatic focus on bridging the gap between theoretical promise and real-world utility. By targeting the fundamental weaknesses of model-based approaches, his work helps pave the way for more robust, sample-efficient learning in robotics. As the field continues to grapple with these challenges, Kveen’s insights offer a valuable step toward making reinforcement learning a viable tool for autonomous systems.