Jongjin Park

About

Jongjin Park is a researcher advancing the frontier of reinforcement learning (RL), with a particular focus on making reward learning more efficient and practical. His key research areas include preference-based RL, semi-supervised learning, and data augmentation techniques for feedback-efficient agent training. Park’s major contribution lies in developing methods that reduce the heavy reliance on human supervision in RL, notably through his work on SURF (Semi-supervised Reward Learning with Data Augmentation for Feedback-efficient Preference-based Reinforcement Learning). This approach, published in 2022 and garnering 14 citations, demonstrates how agents can learn target tasks using only sparse human preferences between behaviors, rather than costly, pre-defined reward functions—a significant step toward scalable, real-world RL applications. By integrating semi-supervised learning and data augmentation, Park’s work addresses a critical bottleneck in preference-based RL, enabling agents to learn effectively from minimal feedback. His research is particularly impactful for students and researchers interested in human-in-the-loop learning, sample efficiency, and the practical deployment of RL systems where reward design is challenging. Park’s contributions continue to shape how we teach agents complex behaviors with less human effort.

Research Focus

Key Achievements

Top Papers

1. 1
   SURF: Semi-supervised Reward Learning with Data Augmentation for Feedback-efficient Preference-based Reinforcement Learning

   14 citations · 2022

Key Collaborators

Contact & Links