Minghan Li
Papers
2
Total Citations
3
H-Index
1
About
Minghan Li is a rising researcher at the intersection of trustworthy AI and embodied intelligence, with key contributions in federated learning fairness and vision-and-language navigation. In healthcare AI, Li introduced **FairFedMed**, a benchmark that systematically evaluates group fairness in federated medical imaging, alongside **FairLoRA**—a parameter-efficient method to mitigate bias across demographic subgroups. This work, already garnering attention with 2 citations since its 2025 release, addresses the critical challenge of ensuring equitable model performance when data is distributed across institutions. Simultaneously, Li pushes the boundaries of spatial reasoning with **NaVid-4D**, a Vision Language Model that unleashes 4D space-time intelligence from egocentric RGB-D videos for Vision-and-Language Navigation. By enabling agents to perceive and act with unprecedented spatial precision, this work (1 citation) overcomes the bottleneck of prior static representations. Li’s dual focus—bridging algorithmic fairness in decentralized healthcare and advancing embodied AI’s spatial cognition—demonstrates a commitment to building both ethical and capable intelligent systems. As these pioneering works gain traction, Li is poised to shape how AI navigates both real-world environments and the complex social landscapes of collaborative learning.
Research Focus
Key Achievements
Top Papers
- 1
- 2