Defending deep neural networks from adversarial attacks on three-dimensional images by compressive sensing

Vladislav Kravets, Bahram Javidi, Adrian Stern

Research output: Contribution to journalConference articlepeer-review

Abstract

We demonstrate the utility of compressive sensing to defend against adversarial attacks on deep learning classifiers and to encrypt the 3D image, thus, to avoid counterattacks.

Original languageEnglish
JournalOptics InfoBase Conference Papers
DOIs
StatePublished - 1 Jan 2021
Event3D Image Acquisition and Display: Technology, Perception and Applications, 3D 2021 - Part of Imaging and Applied Optics Congress 2021 - Virtual, Online, United States
Duration: 19 Jul 202123 Jul 2021

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Mechanics of Materials

Fingerprint

Dive into the research topics of 'Defending deep neural networks from adversarial attacks on three-dimensional images by compressive sensing'. Together they form a unique fingerprint.

Cite this