Block Scrambling Image Encryption Used in Combination with Data Augmentation for Privacy-Preserving DNNs

TL;DR

This paper introduces a learnable image encryption method combining block scrambling and data augmentation to enhance privacy in DNNs while maintaining high classification accuracy.

Contribution

It presents a novel encryption approach that improves robustness against attacks and integrates data augmentation for privacy-preserving deep learning.

Findings

Effective in maintaining classification accuracy with encrypted images

Enhances robustness against various attacks

Demonstrated success in image classification experiments

Abstract

In this paper, we propose a novel learnable image encryption method for privacy-preserving deep neural networks (DNNs). The proposed method is carried out on the basis of block scrambling used in combination with data augmentation techniques such as random cropping, horizontal flip and grid mask. The use of block scrambling enhances robustness against various attacks, and in contrast, the combination with data augmentation enables us to maintain a high classification accuracy even when using encrypted images. In an image classification experiment, the proposed method is demonstrated to be effective in privacy-preserving DNNs.