Jin Hu

I am currently a PhD student (2023-) at Zhongguancun Laboratory (ZGCLab) and State Key Laboratory of Complex & Critical Software Environment, Beihang University, advised by Prof. Xianglong Liu, Prof. Ke Xu and Dr. Jiakai Wang. My research focuses on Physical Adversarial Machine Learning and Visual Generative Modeling, aiming to uncover the dark matter in AI systems and introduce new design paradigms for Trustworthy AI. I plan to graduate around the end of 2027.

In ZGCLab, our group centers on developing physical adversarial examples (PAEs) to evaluate and enhance the safety of vision models for applications like autonomous driving. We will further explore the mechanistic interpretability and scalable applications of PAEs.

News

Feb 10, 2026	One paper published by TPAMI 2026. This work (DynamicPAE) investigates and addresses the training challenges inherent in the end-to-end conditional generation of physical adversarial examples (PAEs). Code is now available at GitHub.
Jan 26, 2026	We are organizing the workshop The 6th Workshop of Adversarial Machine Learning on Computer Vision: Safety of Vision-Language Agents at CVPR 2026 (June 3 or June 4). The workshop will feature Distinguished Paper Awards and competition prizes. We warmly welcome your attention and submissions!

Publications

TPAMI 2026

DynamicPAE: Generating Scene-Aware Physical Adversarial Examples in Real-Time

Jin Hu, Xianglong Liu^*, Jiakai Wang, Junkai Zhang, Xianqi Yang, Haotong Qin, Yuqing Ma, and Ke Xu

IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026

DOI Bib HTML Code

@article{hu2024dynamicpae,
  title = {DynamicPAE: Generating Scene-Aware Physical Adversarial Examples in Real-Time},
  author = {Hu, Jin and Liu, Xianglong and Wang, Jiakai and Zhang, Junkai and Yang, Xianqi and Qin, Haotong and Ma, Yuqing and Xu, Ke},
  selected = true,
  year = {2026},
  volume = {48},
  number = {3},
  pages = {2413-2430},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  doi = {10.1109/TPAMI.2025.3626068},
}

NeurIPS 2025

Exploring Semantic-constrained Adversarial Example with Instruction Uncertainty Reduction

Jin Hu, Jiakai Wang^*, Linna Jing, Haolin Li, Haodong Liu, Haotong Qin, Aishan Liu, Ke Xu, and Xianglong Liu

In Neural Information Processing Systems (NeurIPS), 2025

arXiv Bib Blog Code Website

@inproceedings{Hu2025exploring,
  title = {Exploring Semantic-constrained Adversarial Example with Instruction Uncertainty Reduction},
  author = {Hu, Jin and Wang, Jiakai and Jing, Linna and Li, Haolin and Liu, Haodong and Qin, Haotong and Liu, Aishan and Xu, Ke and Liu, Xianglong},
  booktitle = {Neural Information Processing Systems (NeurIPS)},
  year = {2025},
  selected = true,
  url = {https://nips.cc/virtual/2025/poster/116399},
}

Survey Project

Adversarial Examples in the Physical World: A Survey

Jiakai Wang, Xianglong Liu^*, Jin Hu, Donghua Wang, Siyang Wu, Tingsong Jiang, Yuanfang Guo, Aishan Liu, and Jiantao Zhou

CoRR, 2023

DOI arXiv Bib Website

@article{wang2023AdversarialSurvey,
  title = {Adversarial Examples in the Physical World: {A} Survey},
  author = {Wang, Jiakai and Liu, Xianglong and Hu, Jin and Wang, Donghua and Wu, Siyang and Jiang, Tingsong and Guo, Yuanfang and Liu, Aishan and Zhou, Jiantao},
  journal = {CoRR},
  volume = {abs/2311.01473},
  year = {2023},
  url = {https://doi.org/10.48550/arXiv.2311.01473},
  doi = {10.48550/ARXIV.2311.01473},
  eprinttype = {arXiv},
  eprint = {2311.01473},
  timestamp = {Tue, 07 Nov 2023 18:17:14 +0100},
  biburl = {https://dblp.org/rec/journals/corr/abs-2311-01473.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org},
  selected = true,
}

ICASSP 2025

Generating Targeted Universal Adversarial Perturbation against Automatic Speech Recognition via Phoneme Tailoring

Yujun Zhang, Yanqu Chen, Jiakai Wang^*, Jin Hu, Renshuai Tao, and Xianglong Liu

In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025

Bib HTML

@inproceedings{zhang2025generating,
  title = {Generating Targeted Universal Adversarial Perturbation against Automatic Speech Recognition via Phoneme Tailoring},
  author = {Zhang, Yujun and Chen, Yanqu and Wang, Jiakai and Hu, Jin and Tao, Renshuai and Liu, Xianglong},
  booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages = {1--5},
  year = {2025},
  organization = {IEEE},
  selected = true,
}

Technical report

Evaluating the practicality of learned image compression

Hongjiu Yu, Qiancheng Sun, Jin Hu, Xingyuan Xue, Jixiang Luo, Dailan He, Yilong Li, Pengbo Wang, Yuanyuan Wang, Yaxu Dai, and others

arXiv preprint arXiv:2207.14524, 2022

arXiv Bib

@article{yu2022evaluating,
  title = {Evaluating the practicality of learned image compression},
  author = {Yu, Hongjiu and Sun, Qiancheng and Hu, Jin and Xue, Xingyuan and Luo, Jixiang and He, Dailan and Li, Yilong and Wang, Pengbo and Wang, Yuanyuan and Dai, Yaxu and others},
  journal = {arXiv preprint arXiv:2207.14524},
  selected = true,
  year = {2022}
}