Rank Matters: Understanding and Defending Model Inversion Attacks via Low-Rank Feature Filtering

Accept by KDD 2026.

arXiv: https://arxiv.org/abs/2410.05814

Github: https://github.com/Chrisqcwx/lora_defense

Usage

Please refer to the MIAbox. The scripts are placed in ./scripts

📔 Citation

If you find our work helpful for your research, please kindly cite our papers:

@misc{yu2025rank,
      title={Rank Matters: Understanding and Defending Model Inversion Attacks via Low-Rank Feature Filtering}, 
      author={Hongyao Yu and Yixiang Qiu and Hao Fang and Tianqu Zhuang and Bin Chen and Sijin Yu and Bin Wang and Shu-Tao Xia and Ke Xu},
      year={2025},
      eprint={2410.05814},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      url={https://arxiv.org/abs/2410.05814}, 
}