Tsu-Jui Fu

About

Tsu-Jui Fu is a researcher whose work lies at the intersection of deep reinforcement learning, imitation learning, and self-supervised exploration. In his most-cited paper, "Adversarial Exploration Strategy for Self-Supervised Imitation Learning" (2018), Fu introduced a novel framework that incentivizes an agent to explore its environment without relying on extrinsic rewards or human demonstrations—a significant step toward more autonomous and sample-efficient learning. By pairing a deep reinforcement learning agent with an inverse dynamics model, his approach enables agents to discover meaningful behaviors through intrinsic motivation alone, addressing a core challenge in robotics and AI. Though early in his career, this work has garnered attention for its elegant solution to the exploration problem, accumulating citations that reflect its conceptual impact. Fu’s contributions are particularly relevant for researchers interested in reducing the supervision burden in reinforcement learning, and his adversarial exploration strategy continues to inspire follow-up work in self-supervised and unsupervised policy learning.

Research Focus

Key Achievements

Top Papers

1. 1
   Adversarial Exploration Strategy for Self-Supervised Imitation Learning

   2 citations · 2018

Key Collaborators

Contact & Links