Pattern & Recognition Lab

Yueming Lyu

Assistant Professor
Ph.D. Supervisor
School of Intelligence Science and Technology
Nanjing University

ymlv@nju.edu.cn
Nanjing University Suzhou Campus, No. 1520 Taihu Avenue, Huqiu District, Suzhou, Jiangsu
Yueming Lyu

About Me

I am a tenure-track assistant professor at PRLab (模式识别实验室), School of Intelligence Science and Technology, Nanjing University (Suzhou Campus), directed by Professors Tieniu Tan (谭铁牛) and Caifeng Shan (单彩峰). I earned my PhD from CASIA in 2024, advised by Professor Tieniu Tan (谭铁牛) and co-advised by Professor Jing Dong (董晶). In 2024, I was a visiting researcher at the National University of Singapore.

My research focuses on generative models, large models and agents, and the foundational algorithms of embodied intelligence, together with complementary studies on safety issues that arise when these models are deployed in real-world systems.

In my leisure time, I like playing the guzheng, table tennis, hiking and travelling.

Recent News

2026.05
One paper has been accepted by ICML 2026 (CCF-A).
2026.05
One paper has been accepted by IJCAI 2026 (CCF-B).
2026.02
Two papers have been accepted by CVPR 2026 (CCF-A).
2026.01
Two papers have been accepted by ICLR 2026 (CCF-A).
2026.01
One paper has been accepted by ICASSP 2026 (CCF-B).
2025.12
One paper has been accepted by TIFS 2025 (CCF-A).
2025.10
One paper has been awarded PRCV Best Paper and CCF Outstanding paper.
2025.09
One paper has been accepted by NeruIPS 2025 (CCF-A).
2025.09
One paper has been accepted by WACV 2025.
2025.07
One paper has been accepted by ACM MM 2025 (CCF-A).
2025.01
One paper has been accepted by ICLR 2025 (CCF-A).
2024.11
One paper has been awarded Best Research Paper Award of ICPR MMVPR Workshop.
2024.10
Our team has won the 3rd place in the Multi-Modal Visual Pattern Recognition Challenge # Track 3 @ ICPR2024.
2024.08
I have joined Nanjing University (Suzhou Campus) as an Assistant Professor.
2024.07
One paper has been accepted by ACM MM 2024 (CCF-A).
2024.04
One paper has been accepted by Machine Intelligence Research. (IF=6.4)
2024.04
One paper has been published in TCSVT. (IF=8.4)
2024.01
One paper has been published in TMM. (IF=8.4)
2023.04
One paper has been accepted by CVPR 2023. (CCF-A)
2023.11
One paper has been published in TPAMI. (IF=23.6)
2021.09
One paper has been accepted by ACM MM 2021. (CCF-A)

Research Directions

Content Generation & Editing

I study the architectural design and performance optimization of generative models and world models, and further investigate deployment-related safety issues such as model memorization leakage, privacy protection, and forged-content detection.

Large Models & Agents

I focus on foundation large models, RAG, agent architectures, inference acceleration, and multi-agent collaborative decision-making algorithms, while studying model safety alignment, adversarial attack and defense, and safety evaluation.

Embodied Intelligence

I study perception and decision-making algorithms for embodied intelligence, together with related safety challenges in embodied systems and their deployment scenarios.

AI Safety & Security

I work on building trustworthy AI systems for making network information credible, including misinformation detection, content provenance, robustness assessment, and security mechanisms for generative and agentic models, as part of NICE-PRLab @ NJU.

Join Us

We welcome highly motivated students with strong research interests and relevant experience. Applicants with first-author CCF-A/B publications or academic competition awards are especially encouraged to contact us. The group continuously recruits research interns at undergraduate, master's, Ph.D., and gap-year stages, as well as postdoctoral researchers.

2027 Graduate / Ph.D. Students
  • Multiple openings for the 2027 cohort
  • Backgrounds in computer science, AI, automation, or related fields are welcome
  • Strong foundations in programming, mathematics, and deep learning
  • High-quality publications or competition experience are preferred
  • Long-term research motivation and self-direction
Research Directions
  • Generation and editing: Video World Models and World Action Models
  • Agent systems: memory, multi-agent collaboration, and decision interaction
  • Model-intrinsic safety: jailbreak defense, backdoor detection, and prompt injection defense
  • Synthetic image and video detection, provenance, and attribution
  • Multimodal large model safety and ethics
Research Interns / Postdocs
  • Undergraduate, master's, Ph.D., and gap-year students are welcome
  • Postdoctoral researchers are recruited on a rolling basis
  • Students with clear research interests are encouraged to reach out
  • Opportunities to collaborate with academic and industry partners
  • Future Ph.D. applicants are especially welcome

Interested in joining us?

Please send your CV by email.