Junfei Wu

Institute of Automation, Chinese Academy of Sciences

Beijing, China

junfei.wu@cripac.ia.ac.cn

I’m Junfei Wu, a Ph.D. student at the Institute of Automation, Chinese Academy of Sciences (State Key Laboratory of Multimodal Artificial Intelligence), advised by Prof. Tieniu Tan. I also collaborate closely with Prof. Qiang Liu, Shu Wu, and Liang Wang. My research centers on trustworthy and interpretable reasoning in multimodal large language models (MLLMs), with a focus on spatial reasoning, hallucination detection, and test-time adaptation.

I interned at Ant Group’s Multimodal LLM team, where I developed the “Drawing to Reason in Space” framework to enhance visual thinking in LVLMs—published at NeurIPS 2025. I also pioneered representation-based hallucination mitigation techniques (EMNLP 2025) and logic-consistency-based detection frameworks (ACL 2024 Findings).

When not debugging tensors, I enjoy playing basketball, badminton, table tennis, yo-yo, or solving a Rubik’s Cube.

Feel free to reach me at junfei.wu@cripac.ia.ac.cn.

news

Sep 19, 2025	Our paper “Reinforcing Spatial Reasoning in Vision-Language Models with Interwoven Thinking and Visual Drawing” has been accepted to the NeurIPS 2025. See you in San Diego!
Aug 21, 2025	Our paper “SHARP: Steering Hallucination in LVLMs via Representation Engineering” has been accepted to EMNLP 2025. See you in Suzhou!

selected publications

Reinforcing spatial reasoning in vision-language models with interwoven thinking and visual drawing

Junfei Wu, Jian Guan, Kaituo Feng, and 5 more authors

arXiv preprint arXiv:2506.09965, 2025

Bib

@article{wu2025reinforcing,
  title = {Reinforcing spatial reasoning in vision-language models with interwoven thinking and visual drawing},
  author = {Wu, Junfei and Guan, Jian and Feng, Kaituo and Liu, Qiang and Wu, Shu and Wang, Liang and Wu, Wei and Tan, Tieniu},
  journal = {arXiv preprint arXiv:2506.09965},
  year = {2025},
}

SHARP: Steering Hallucination in LVLMs via Representation Engineering

Junfei Wu, Yue Ding, Guofan Liu, and 7 more authors

In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, 2025

Bib

@inproceedings{wu2025sharp,
  title = {SHARP: Steering Hallucination in LVLMs via Representation Engineering},
  author = {Wu, Junfei and Ding, Yue and Liu, Guofan and Xia, Tianze and Huang, Ziyue and Sui, Dianbo and Liu, Qiang and Wu, Shu and Wang, Liang and Tan, Tieniu},
  booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  pages = {14357--14372},
  year = {2025},
}

Logical Closed Loop: Uncovering Object Hallucinations in Large Vision-Language Models

Junfei Wu, Qiang Liu, Ding Wang, and 4 more authors

In Findings of the Association for Computational Linguistics ACL 2024, 2024

Bib

@inproceedings{wu2024logical,
  title = {Logical Closed Loop: Uncovering Object Hallucinations in Large Vision-Language Models},
  author = {Wu, Junfei and Liu, Qiang and Wang, Ding and Zhang, Jinghao and Wu, Shu and Wang, Liang and Tan, Tieniu},
  booktitle = {Findings of the Association for Computational Linguistics ACL 2024},
  pages = {6944--6962},
  year = {2024},
}

Adversarial contrastive learning for evidence-aware fake news detection with graph neural networks

Junfei Wu, Weizhi Xu, Qiang Liu, and 2 more authors

IEEE Transactions on Knowledge and Data Engineering, 2023

Bib

@article{wu2023adversarial,
  title = {Adversarial contrastive learning for evidence-aware fake news detection with graph neural networks},
  author = {Wu, Junfei and Xu, Weizhi and Liu, Qiang and Wu, Shu and Wang, Liang},
  journal = {IEEE Transactions on Knowledge and Data Engineering},
  volume = {36},
  number = {11},
  pages = {5591--5604},
  year = {2023},
  publisher = {IEEE},
}

Bias mitigation for evidence-aware fake news detection by causal intervention

Junfei Wu, Qiang Liu, Weizhi Xu, and 1 more author

In Proceedings of the 45th International ACM SIGIR conference on research and development in information retrieval, 2022

Bib

@inproceedings{wu2022bias,
  title = {Bias mitigation for evidence-aware fake news detection by causal intervention},
  author = {Wu, Junfei and Liu, Qiang and Xu, Weizhi and Wu, Shu},
  booktitle = {Proceedings of the 45th International ACM SIGIR conference on research and development in information retrieval},
  pages = {2308--2313},
  year = {2022},
}