Yuji Shi
Papers
1
Total Citations
19
H-Index
1
About
Yuji Shi is a researcher advancing the frontiers of intelligent robotics and machine perception, with a primary focus on unsupervised domain adaptation (UDA) and knowledge distillation. Their most cited work, "Model Adaptation through Hypothesis Transfer with Gradual Knowledge Distillation" (2021, 19 citations), tackles a fundamental challenge in robotics: enabling perception systems to adapt autonomously to changing environments without labeled data. Shi’s major contribution lies in formulating domain adaptation as a hypothesis transfer problem, where knowledge is gradually distilled from a source model to a target domain, preserving performance while reducing catastrophic forgetting. This approach bridges the gap between theoretical transfer learning and practical robotic deployment, offering a scalable solution for real-world scenarios like autonomous navigation or manipulation in novel settings. By integrating gradual distillation with hypothesis transfer, Shi’s work has influenced subsequent research on efficient, label-free adaptation, earning recognition among peers for its clarity and applicability. Their research continues to push toward more robust, adaptable AI systems, making Shi a notable voice in the intersection of robotics and machine learning.
Research Focus
Key Achievements
Top Papers
- 1