Residue-Net: Multiplication-free Neural Network by In-situ, No-loss Migration to Residue Number Systems

Guo, Y.; Imani, M.; Kang, J.; Salamat, S.; Morris, J.; Aksanli, B.; Kim, Y. and

Abstract Computer applications often leave traces or residues that enable forensic examiners to gain a detailed understanding of the actions a user performed on a computer. Such digital breadcrumbs are left by a large variety of applications, potentially (and indeed likely) unbeknownst to their users. This paper presents the concept of residue-free computing in which a user can operate any existing application installed on their computer in a mode that prevents trace data from being recorded to disk, thus frustrating the forensic process and enabling more privacy-preserving computing. In essence, residue-free computing provides an “incognito mode” for any application. We introduce our implementation of residue-free computing, R esidue F ree , and motivate R esidue F ree by inventorying the potentially sensitive and privacy-invasive residue left by popular applications. We demonstrate that R esidue F ree allows users to operate these applications without leaving trace data, while incurring modest performance overheads.

More Like this