Authors: Dingfan Chen, Marie Oestreich, Tejumade Afonja, Raouf Kerkouche, Matthias Becker, Mario Fritz

Published on: February 07, 2024

Impact Score: 8.05

Arxiv code: Arxiv:2402.04912

Summary

• What is new: Systematic analysis of DP generative models on real-world gene expression data, uncovering limitations in capturing biological characteristics.
• Why this is important: Existing DP generative models focus on basic datasets and show promise only in simple metrics, not addressing complex data like gene expression.
• What the research proposes: Conducted a comprehensive analysis of five DP generation methods focusing on their application in real-world gene expression data.
• Results: While methods show reasonable utility, none accurately capture the biological specifics of the data, indicating a need for model improvements.

Technical Details

Technological frameworks used: Differential Privacy

Models used: Five representative DP generation methods

Data used: Real-world gene expression data

Potential Impact

Biotech companies and healthcare sectors focusing on gene analysis and synthetic data generation.

Want to implement this idea in a business?

We have generated a startup concept here: GeneGuard.