I am currently a Computer Science master’s student at Yale University. I earned my bachelor’s degree in Computer Science from Sun Yat-sen University, where I was advised by Changdong Wang and Guanbin Li. During my undergraduate studies, my research focused on two areas: (1)AI Security and Privacy, with an emphasis on federated learning; (2) Vision Models for Robotic Surgery Assistance, including segmentation and video understanding tasks.
Currently, my research interests lie in Large Language Models (LLMs) and Multi-modal Large Language Models (MLLMs), with a focus on their adaptation and applications in real-world scenarios.
🔥 News
- 2026.02: 🎉🎉 Our work “SurgVidLM: Towards Multi-grained Video Understanding with Large Language Model in Robot-assisted Surgery” is accepted by ICRA 2026!
- 2025.06: 🎉🎉 Our work “Find a Scapegoat: Poisoning Membership Inference Attack and Defense to Federated Learning” is accepted by ICCV 2025!
- 2025.06: 🎉🎉 Our work “PDZSeg: Adapting the Foundation Model for Dissection Zone Segmentation with Visual Prompts in Robot-assisted Endoscopic Submucosal Dissection” won The IHU Strasbourg and NDI Bench to Bedside Award at IPCAI 2025!
- 2025.05: 🎉🎉 Our work “ETSM: Automating Dissection Trajectory Suggestion and Confidence Map-Based Safety Margin Prediction for Robot-assisted Endoscopic Submucosal Dissection” won the Best Application Award at the Multi-Scale Medical Robotics Center(MRC) Symposium 2025!
- 2025.02: 🎉🎉 Our work “PDZSeg: Adapting the Foundation Model for Dissection Zone Segmentation with Visual Prompts in Robot-assisted Endoscopic Submucosal Dissection” is accepted by IPCAI 2025!
- 2025.01: 🎉🎉 Our work “ETSM: Automating Dissection Trajectory Suggestion and Confidence Map-Based Safety Margin Prediction for Robot-assisted Endoscopic Submucosal Dissection” is accepted by ICRA 2025!
📝 Publications
Guankun Wang*, Junyi Wang, Wenjin Mo, et al.
- Dataset: Built SVU-31K with 31K+ video-instruction pairs for both global and fine-grained surgical video understanding.
- Model: Developed SurgVidLM with two-stage global-local video reasoning and Multifrequency Fusion Attention to preserve task-specific details across temporal scales.
Find a Scapegoat: Poisoning Membership Inference Attack and Defense to Federated Learning
Wenjin Mo, Zhiyuan Li, Minghong Fang, Mingwei Fang
- Proposed an innovative Poisoning Membership Inference Attack (PMIA) for federated learning, formulated from the perspective of optimizing the angular deviation of malicious updates.
- Developed a novel Byzantine-robust defense mechanism against the proposed PMIA by introducing an angle-based filtering criterion.
Mengya Xu, Wenjin Mo, Guankun Wang, Huxin Gao, An Wang, Long Bai, Chaoyang Lyu, Xiaoxiao Yang, Zhen Li, Hongliang Ren
- Introduced a framework that combines optimal dissection trajectory prediction with a confidence map-based safety margin, providing a more secure and intelligent decision-making tool to minimize surgical risks for ESD procedures.
Mengya Xu, Wenjin Mo, Guankun Wang, Huxin Gao, An Wang, Zhen Li, Xiaoxiao Yang, Hongliang Ren
- Introduced a prompted-based dissection zone segmentation (PDZSeg) model, which is specifically designed for dissection zone segmentation and can effectively utilize various visual prompts, including scribbles and bounding boxes.
🎖 Honors and Awards
- 2025.06 Outstanding Graduate Student
- 2024.10 University Second-Class Scholarship, 2023 (top 10%)
- 2023.12 Meritorious Award-Mathematical Contest in Modeling (top 10%)
- 2023.10 University First-Class Scholarship, 2022 (top 5%)
- 2022.12 Provincial First Prize, China Undergraduate Mathematical Contest in Modeling (top 10%)
- 2022.10 University First-Class Scholarship, 2021 (top 5%)
📖 Educations
- 2021.09 - 2025.06, B.Eng., Computer Science and Technology, Sun Yat-sen University, Guangzhou, China
💻 Internships
- 2025.04 - 2025.12, Multi-modal Large Language Models (MLLMs) Engineer, Alibaba, China.