Lijun Sheng (生力军)

I am currently a Ph.D. student in University of Science and Technology of China, supervised by Prof. Tieniu Tan and jointly supervised by Prof. Jian Liang, Prof. Ran He and Prof. Zilei Wang. Before that, I received bachelor’s degree in Automation in University of Science and Technology of China in 2020.
My research interests are unsupervised learning during test time, trustworthy machine learning, and visual language models.
news
Sep 19, 2025 | A paper is accepted by NeurIPS 2025 (D&B Track). |
---|---|
Feb 27, 2025 | A paper is accepted by CVPR 2025. |
Dec 10, 2024 | A paper is accepted by AAAI 2025. |
Sep 30, 2023 | Going to Paris to attend ICCV 2023 ! |
Jul 14, 2023 | A paper is accepted by ICCV 2023. |
publications
- ICCVAdaptGuard: Defending Against Universal Attacks for Model AdaptationIn International Conference on Computer Vision (ICCV) , 2023
- AAAIProtecting Model Adaptation from Trojans in the Unlabeled DataIn AAAI Conference on Artificial Intelligence (AAAI) , 2025
- CVPRR-TPT: Improving Adversarial Robustness of Vision-Language Models through Test-Time Prompt TuningIn IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2025
- NeurIPSThe Illusion of Progress? A Critical Look at Test-Time Adaptation for Vision-Language ModelsIn NeurIPS Datasets and Benchmarks Track , 2025
- Adapting Vision-Language Models Without Labels: A Comprehensive SurveyarXiv preprint arXiv:2508.05547, 2025
- ICMLRealistic Unsupervised CLIP Fine-tuning with Universal Entropy OptimizationIn International Conference on Machine Learning (ICML), Spotlight , 2024
- NNProxyMix: Proxy-based Mixup Training with Label Refinery for Source-Free Domain AdaptationNeural Networks, 2023