Exploring Image Reconstruction Attack in Deep Learning Computation Offloading

摘要: Deep learning (DL) computation offloading is commonly adopted to enable the use of computation-intensive DL techniques on resource-constrained devices. However, sending private user data an external server raises a serious privacy concern. In this paper, we introduce privacy-invading input reconstruction method which utilizes intermediate pipeline. doing so, first define Peak Signal-to-Noise Ratio (PSNR)-based metric for assessing quality. Then, simulate attack diverse models find out relationship between model structures and performance attacks. Finally, provide several insights structure design prevent reconstruction-based attacks: using skip-connection, making deeper, including various operations such as inception module.