# Privacy-Preserving Deep Neural Networks with Pixel-based Image Encryption Considering Data Augmentation in the Encrypted Domain

## TL;DR

This paper introduces a new pixel-based image encryption method for privacy-preserving deep neural networks, enabling data augmentation in the encrypted domain and improving classification accuracy over existing methods.

## Contribution

It proposes a novel pixel-based encryption technique and an adaptation network that together allow data augmentation in encrypted images for DNNs.

## Key findings

- The method enables data augmentation in the encrypted domain.
- It outperforms existing privacy-preserving methods in classification accuracy.
- The approach is effective with ResNet-18 for image classification.

## Abstract

We present a novel privacy-preserving scheme for deep neural networks (DNNs) that enables us not to only apply images without visual information to DNNs for both training and testing but to also consider data augmentation in the encrypted domain for the first time. In this paper, a novel pixel-based image encryption method is first proposed for privacy-preserving DNNs. In addition, a novel adaptation network is considered that reduces the influence of image encryption. In an experiment, the proposed method is applied to a well-known network, ResNet-18, for image classification. The experimental results demonstrate that conventional privacy-preserving machine learning methods including the state-of-the-arts cannot be applied to data augmentation in the encrypted domain and that the proposed method outperforms them in terms of classification accuracy.