Privacy-Preserving Remote Sensing Image Generation and Classification with Differentially Private GANs

• Published in: IEEE sensors journal 2023-09, Vol.23 (18), p.1-1
• Main Authors: Huang, Yujian, Cao, Lei
• Format: Article
• Language: English
• Subjects: Data models; Differential privacy; Generative adversarial networks; Generative Adversarial Networks (GANs); Hyperspectral Image Classification; Hyperspectral imaging; Image classification; Image processing; Image quality; Model accuracy; Privacy; Privacy-Preserving Machine Learning; Remote sensing; Sensors; Synthetic data; Training
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2023.3267001

Generative Adversarial Networks (GANs) have demonstrated their remarkable capacity to learn the training data distribution and produce high-quality synthetic images, which have been widely adopted in image recognition tasks in remote sensing research communities. However, previous work has shown that using GANs does not preserves privacy, e.g., being susceptible to membership attacks, while sensitive information is vulnerable to nefarious activities. This drawback is considered severe in remote sensing communities, in which critical researches highly value the security and privacy of the image content. Thus, to publicly share sensitive data for supporting critical researches, in the meantime guarantee the model accuracy trained from privacy-preserving data, this work develops GANs within the Differential Privacy (DP) framework, and proposes a Remote Sensing Differentially Private Generative Adversarial Networks (RS-DPGANs) for both privacy-preserving synthetic image generation and classification. Our RS-DPGANs is capable of releasing safe-version of synthetic data meanwhile obtaining favorable classification r esults, w hich gives rigorous guarantees for the privacy of sensitive data and balance between the model accuracy and privacy-preserving degree. Extensive empirical and statistical results both confirm the effectiveness of our framework.