Ziwen Li

Ziwen Li

Southwest University

I am a third-year undergraduate student majoring in Software Engineering at the School of Computer and Information Science, Southwest University. My research interests focus on generative vision and large language models, with particular emphasis on diffusion-based image generation and panoramic/scene generation. I am currently supervised by Prof. Xianfeng Han (Google Scholar) and am actively seeking opportunities for MPhil programs starting in Fall 2027. Please feel free to contact me directly via email.

I am currently working as a research intern at The Chinese University of Hong Kong, Shenzhen.

🔥 News

  • 2026.02:  📝 GeoPano: Geometry-Aware Diffusion for Text to 360-degree Panoramas (first author) — under review at ACM MM (CCF-A).
  • 2026.02:  📝 NGM-RAG: Neural Graph Matching based Retrieval-Augmented Generation (co-second author) — under review at ESWA (JCR Q1).
  • 2026.02:  📝 When Retrieval is Not Enough: Uncertainty-Driven Attacks to Retrieval-Augmented Generation System (co-author) — under review at CCS (CCF-A).

    📝 Publications

ACM MM (CCF-A)
sym

GeoPano: Geometry-Aware Diffusion for Text to 360-degree Panoramas Under Review

Ziwen Li, Xueyi Zhang, Chengwei Zhang, Xian-Feng Han, Jinlin Guo .

Project

  • We study diffusion-based panorama generation and introduce geometry-aware constraints with reinforcement learning-based fine-tuning to reduce boundary discontinuity and distortion in 360° views, achieving competitive results on public panorama benchmarks.
ESWA (JCR Q1)
sym

NGM-RAG: Neural Graph Matching based Retrieval-Augmented Generation Under Review

Guo Chen, Ziwen Li, Maolin Zheng, Qi Cao, Junjie Huang, Tao Jia.

Project

  • We study diffusion-based panorama generation and introduce geometry-aware constraints with reinforcement learning-based fine-tuning to reduce boundary discontinuity and distortion in 360° views, achieving competitive results on public panorama benchmarks.
CCS (CCF-A)
sym

When Retrieval is Not Enough: Uncertainty-Driven Attacks to Retrieval-Augmented Generation System Under Review

Guo Chen, Junjie Huang, Yulei Long, Zixuan Huang, Yilan Huang, Ziwen Li, Qi Cao, Hao Gao, Huawei Shen.

Project

  • We reveal that hallucination in RAG is an inherent result of decision-making under epistemic uncertainty, and propose PhantomRAG — a lightweight poisoning attack that subtly steers model outputs by injecting benign-looking documents to manipulate uncertainty resolution, even when retrieval remains semantically correct.

🎖 Honors and Awards

  • 2025 Provincial First Prize, National Undergraduate Mathematical Contest in Modeling (Higher Education Press Cup), China.
  • 2025 Honorable Mention (H Award), Mathematical Contest in Modeling / Interdisciplinary Contest in Modeling (MCM/ICM), International.
  • 2025 Finalist (F Award), Certified Cup International Mathematical Modeling Competition, China.
  • 2025 Second-Class Scholarship, Southwest University (University Level).
  • 2024 Third-Class Scholarship, Southwest University (University Level).

📖 Educations

  • Sep. 2023 - Present, B.Eng. in Software Engineering, Southwest University, Chongqing, China. (GPA: 3.67 / 5.0)
  • Mar. 2026 - Present, Research Intern, The Chinese University of Hong Kong, Shenzhen, Shenzhen, China.

    💻 Internships

  • Jun. 2024 - Sep. 2024, Chongqing CAS Automotive Software Innovation Center, Chongqing, China.
    Project Leader — Indoor autonomous navigation modeling with IsaacSim & ROS; Gazebo simulation and 2D SLAM with Gmapping to improve navigation accuracy in complex indoor environments.