Hi! Currently, I’m a researcher at M17 focusing on MoE architecture, long-horizon agent & LLM, and unified understanding & generation. I received my Ph.D. from Shanghai Jiao Tong University at 2024.06, and fortunately to be supervised by Prof. Junchi Yan. My doctoral research primarily focuses on open-set recognition, and prompt learning on vision/multimodal foundation models. I am honored to have collaborated closely with research teams during my internships at Alibaba (working with Dr. Guoxin Wang and Dr. Chenyi Lei), Huawei Cloud (working with Dr. Qi Tian and Dr. Xiaopeng Zhang), and Shanghai AI Laboratory (working with Prof. Junchi Yan). Prior to this, I obtained my bachelor’s degree (1/106) from Xidian University at 2019.06.

News

• [2026.06] I will serve as the Area Chair for the 2nd MUGG workshop at ECCV 2026.
• [2026.06] EvoTok（Co-First & Project Lead）and GRADE are accepted to ECCV 2026.🎉🎉🎉
• [2026.03] We opensource the FineRMoE: preprint, model and code based on Megatron-LM framework.
• [2026.02] Two paper about video tokenization and object detection are accepted to CVPR 2026. 🎉🎉🎉
• [2026.01] One paper about token selection is accepted to ICLR 2026. 🎉🎉🎉
• [2026.01] One paper about open vocabulary object detection is accepted to IJCV. 🎉🎉🎉
• [2025.12] I give a talk about MoE upcycling and architecture innovation at CCLD 2025.
• [2025.09] One paper about visual prompt learning is accepted to Artificial Intelligence. 🎉🎉🎉
• [2025.09] One paper about multi-task learning is accepted to NeurIPS 2025. 🎉🎉🎉
• [2025.07] We release the scientific foundation model Innovator at WAIC 2025.
• …

Selected Publications

(*Equal Contribution, †Corresponding Author/Project Lead)

• ACE: Pluggable Adaptive Context Elasticizer across Agents [Preprint]
  Ning Liao, Zihao Long, Xiaoxing Wang, et al.
• FineRMoE: Dimension Expansion for Finer-Grained Expert with Its Upcycling Approach [Preprint][Code][Model]
  Ning Liao, Xiaoxing Wang, Xiaohan Qin, Junchi Yan
• Innovator: Scientific Continued Pretraining with Fine-grained MoE Upcycling [Tech Report at WAIC 2025]
  Ning Liao, Xiaoxing Wang, Zehao Lin, et al.
• CoHOZ: Contrastive Multimodal Prompt Tuning for Hierarchical Open-set Zero-shot Recognition [ACM MM 2022 Oral]
  Ning Liao, Yifeng Liu, Xiaobo Li, Chenyi Lei, Guoxin Wang, Xian-Sheng Hua, Junchi Yan
• Visual Prompt Learning: A Survey [Chinese Journal of Computers]
  Ning Liao, Min Cao, Junchi Yan
• M-Tuning: Prompt Tuning with Mitigated Label Bias in Open-Set Scenarios [IEEE TCSVT]
  Ning Liao, Xiaopeng Zhang, Min Cao, Junchi Yan
• Rethinking visual prompt learning as masked visual token modeling [Artificial Intelligence]
  Ning Liao†, Bowen Shi, Xiaopeng Zhang, Min Cao, Junchi Yan, Qi Tian
• EvoTok: A Unified Image Tokenizer via Residual Latent Evolution for Visual Understanding and Generation [ECCV 2026]
  Yan Li*, Ning Liao*, Xiangyu Zhao, et al.
• AutoAgent: Evolving Cognition and Elastic Memory Orchestration for Adaptive Agents [Preprint]
  Xiaoxing Wang*, Ning Liao*(dominating Elastic Memory Orchestration), Shikun Wei, et al.
• Toward Open Vocabulary Aerial Object Detection with CLIP-Activated Student-Teacher Learning [ECCV 2024]
  Yan Li, Weiwei Guo, Xue Yang, Ning Liao, Dunyun He, Jiaqi Zhou, Wenxian Yu
• Exploiting Unlabeled Data with Multiple Expert Teachers for Open Vocabulary Aerial Object Detection and Its Orientation Adaptation [IJCV] Yan Li, Weiwei Guo, Xue Yang, Ning Liao, Shaofeng Zhang, Yi Yu, Wenxian Yu, Junchi Yan
• Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space [CVPR 2026]
  Yan Li, Changyao Tian, Renqiu Xia, Ning Liao, Weiwei Guo, Junchi Yan, Hongsheng Li, Jifeng Dai, Hao Li, Xue Yang

Academic Services

Reviewer: CVPR, ICLR, ICML, AAAI, IEEE TMM, IEEE TCSVT.