Hello! 👋 I am an incoming MPhil student at CUHK-SZ, fortunately advised by Prof. Xiaoying Tang. Prior to this, I received my B.Eng. in Electronic and Information Engineering from UESTC. My research broadly focuses on LVLMs. Specifically, I am interested in Trustworthy AI, Safety Alignment, and LVLM SFT/RL. Earlier, I explored Industrial LVLMs (wafer defect diagnosis) and cross-domain fault diagnosis.
⭐ I am currently in a gap period and actively seeking research internship opportunities (industry/academia) in LVLMs or AI safety/alignment. I am available immediately for onsite roles! Please feel free to contact me via email: ChenxiLi_lcx@outlook.com or WeChat: chenxi_yef. I warmly welcome potential collaborations and look forward to connecting!
🔥 News
- 2026.03: 📝 Submitted one first author paper on LVLM hallucination mitigation to ECCV 2026 (currently under review).
- 2026.02: 🎉 One first author paper on black-box jailbreaking for LVLMs was accepted to CVPR 2026! This work was completed during my second internship for AI safety research.
- 2025.07: 🎉 Invited to serve as a reviewer for AAAI 2026.
- 2025.06: 🚀 Started my journey at DAIL Tech as a Research Intern. The AI startup is founded by my peer Xiande Huang.
- 2025.06: 🎓 Graduated from UESTC with a First-Class Honours degree (joint program with UoG).
- 2024.11: 🏆 Awarded the Academic Scholarship (Top 5%, 30,000 RMB).
- 2024.07: 🎉 One co-first author paper accepted to IEEE Transactions on Industrial Informatics (SCI Q1), completed during my first research internship in industrial LVLMs for smart manufacturing.
- 2024.04: 🎉 One first author paper accepted to IEEE Transactions on Reliability (SCI Q1), completed during my first research internship in transfer learning for smart manufacturing, and underwent a year of review.
- 2022.10: 🌱 The starting point of my academic journey! Joined Tsinghua University as a research assistant.
📝 Publications & Preprints
Models as Lego Builders: Assembling Malice from Benign Blocks via Semantic Blueprints
Chenxi Li*, Xianggan Liu*, Dake Shen, Yaosong Du, Zhibo Yao, Hao Jiang, Linyi Jiang, Chengwei Cao, Jingzhe Zhang, RanYi Peng, Peiling Bai, Xiande Huang
- 🔥 TL;DR: Proposed StructAttack, a black-box one-shot jailbreak framework utilizing semantic slot filling, achieving highly efficient attacks against LVLMs with minimal overhead.
MAP: Mitigating Hallucinations in Large Vision-Language Models with Map-Level Attention Processing
Chenxi Li*, Yichen Guo*, Benfang Qian, Jinhao You, Kai Tang, Yaosong Du, Zonghao Zhang, Xiande Huang
- 🔥 TL;DR: Proposed MAP, a training-free decoding framework that treats hidden states as a 2D semantic map, utilizing Layer-Wise Criss-Cross Attention and logit fusion to effectively mitigate LVLM hallucinations.
Chenxi Li*, Huan Wang*, Yan-Fu Li
- 🔥 TL;DR: Proposed DefectGLM, the first LVLM tailored for industrial visual monitoring. It leverages a novel multimodal wafer dataset, LoRA-based contrast visual adaptation, and instruction tuning for highly accurate defect diagnosis.
Chenxi Li, Huan Wang, Te Han
- 🔥 TL;DR: Proposed DSPC-A, an end-to-end framework that integrates an auxiliary network to learn clean label distributions from noisy pseudo-labels, enabling precise subdomain alignment for cross-domain fault diagnosis.
Huan Wang, Chenxi Li, Yan-Fu Li, Fugee Tsung
- 🔥 TL;DR: Proposed IVMMF, a comprehensive framework integrating large visual models for defect identification and local-knowledge-augmented LLMs to provide professional maintenance suggestions for engineers.
AIVA: An AI-based Virtual Companion for Emotion-aware Interaction
Chenxi Li
- 🔥 TL;DR: Proposed AIVA, an AI virtual companion framework integrating a Multimodal Sentiment Perception Network (MSPN) and emotion-aware prompt engineering to enable empathetic and emotionally aligned human-computer interaction.
🎖 Honors and Awards
- [2025] Honors Research Degree (awarded to exceptional graduates for research excellence)
- [2024] Academic Scholarship (Top 5%, 30,000 RMB)
- [2024] James Watt Innovation Scholarship
- [2024] First-Class Scholarship for Outstanding Students (Top 10%)
📖 Educations
- 2026.09 - 2028.06 (Expected), MPhil in Computer and Information Engineering, School of Science and Engineering, The Chinese University of Hong Kong, Shenzhen (CUHK-SZ)
- 2021.09 - 2025.06, B.Eng. in Electronic and Information Engineering, Glasgow College, University of Electronic Science and Technology of China (UESTC)
💻 Internships
- 2025.06 - 2026.02, Research Intern, [DAIL Tech], China.
- 2022.10 - 2024.04, Research Assistant, [Tsinghua University], China.