Summary of Mass: Multi-attribute Selective Suppression For Utility-preserving Data Transformation From An Information-theoretic Perspective, by Yizhuo Chen et al.
MaSS: Multi-attribute Selective Suppression for Utility-preserving Data Transformation from an Information-theoretic Perspective
by Yizhuo Chen, Chun-Fu Chen, Hsiang Hsu, Shaohan Hu, Marco Pistoia, Tarek Abdelzaher
First submitted to arxiv on: 23 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 The proposed framework for utility-preserving privacy protection in large-scale datasets tackles the limitations of existing methods by providing a formal information-theoretic definition and a learnable data transformation approach. The framework selectively suppresses sensitive attributes while preserving useful ones, regardless of whether they are known or annotated. Rigorous theoretical analyses demonstrate operational bounds, and comprehensive experiments on various modalities (facial images, voice audio, human activity motion sensor signals) show the effectiveness and generalizability of the method across different configurations and tasks. The code is available for public use. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research paper proposes a new way to protect people’s private information in large datasets while still keeping the useful data. Right now, there are many ways to do this, but they have problems like making the data less helpful or not being based on solid ideas. To fix these issues, the authors came up with a new method that uses math and computer learning to change the data so that sensitive information is hidden while important details remain intact. They tested their approach on different types of data and showed it works well. |
