Summary of Adaptive Hybrid Masking Strategy For Privacy-preserving Face Recognition Against Model Inversion Attack, by Yinggui Wang et al.
Adaptive Hybrid Masking Strategy for Privacy-Preserving Face Recognition Against Model Inversion Attack
by Yinggui Wang, Yuanqing Huang, Jianshu Li, Le Yang, Kai Song, Lei Wang
First submitted to arxiv on: 14 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces an adaptive hybrid masking algorithm to protect personal sensitive data used in training face recognition (FR) models from model inversion attacks (MIA). The algorithm utilizes frequency domain mixing with an enhanced adaptive MixUp strategy based on reinforcement learning. This approach allows for a larger number of images to be mixed while maintaining satisfactory recognition accuracy. The proposed method optimizes privacy protection by maximizing the reward function during training, striking a balance between privacy and accuracy. Experimental results show that the hybrid masking scheme outperforms existing defense algorithms in terms of privacy preservation and recognition accuracy against MIA. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps keep our faces safe online! When we use face recognition technology, bad guys can try to figure out what our original training data looks like. That’s not cool! To stop this from happening, the researchers developed a new way to mix up face images in the frequency domain. This makes it harder for the bad guys to get information they shouldn’t have. The new method is really good at balancing privacy and accuracy, so we can use face recognition tech without worrying about our personal info being stolen. |
Keywords
* Artificial intelligence * Face recognition * Reinforcement learning
