Fanyu Zhao
赵梵宇
I am a second-year Master's student at the School of Computer Science and Technology, Fudan University, advised by Assoc. Prof. Yinsheng Li. I received my B.S. in Pharmacy from Fudan University in 2024. My research interests include AI Agents (tool use, MCP, multi-agent systems, evaluation), few-shot 3D point cloud segmentation, and AI for science (e.g., GNNs for molecular/crystal tasks).
News
- [May 2026] I presented our paper Uncertainty-aware Prototype Learning with Variational Inference for Few-shot Point Cloud Segmentation as an oral talk at ICASSP 2026 in Barcelona, Spain.
- [Mar 2026] Our paper HOP3D is accepted to ICME 2026.
- [Jan 2026] Our paper on uncertainty-aware prototype learning for few-shot point cloud segmentation is accepted to ICASSP 2026 (Oral).
Publications
* Co-first author
Accepted / Published
Uncertainty-aware Prototype Learning with Variational Inference for Few-shot Point Cloud Segmentation
Y. Zhao*, F. Zhao*, Y. Li. ICASSP 2026 (Oral).
HOP3D: Learning Hierarchical Orthogonal Prototypes for Generalized Few-Shot 3D Point Cloud Segmentation
Y. Zhao*, F. Zhao*, Z. Zhang, S. Wu, Y. Lin, Y. Li. ICME 2026.
Population Pharmacokinetics and Thrombocytopenia Risk Assessment of Linezolid in Liver Transplant Recipients
X. Shi*, W. Yang*, F. Zhao, et al. Journal of Antimicrobial Chemotherapy, 2025.
Under Review / In Preparation
Uncertainty Gates the Gradient: Adaptive Updates for Generalized Few-Shot 3D Segmentation
Y. Zhao*, F. Zhao*, S. Wu, H. Zhang, R. Cao, Y. Li. NeurIPS 2026 (under review).
MCC-GCN: Leveraging Graph Convolutional Networks for Multicomponent Crystal Classification
Y. Deng, F. Zhao, X. Zhou, et al. Nature Methods (under review).
AMQ-Bench: Agent-Oriented MCP Quality Benchmark
F. Zhao, Y. Zhao, H. Zhang, Y. Lin, S. Wu, R. Cao, Y. Li. NeurIPS 2026 Datasets and Benchmarks Track (under review).
Projects
Micro-Agent — Lightweight AI agent framework for vertical domains (ReAct, RAG, MCP, skills). Used in national key R&D deployments; demo: fdueblab.cn.
AMQ-Bench — Benchmark and evaluation pipeline for agent-facing MCP service quality (Availability / Usability / Utility).
GenericAgent — Open general-purpose agent; I contributed to execution-efficiency optimizations during my internship.