Type

Conference Proceedings

Authors

Kevin McGuinness
Suzanne Little
Panagiotis Linardos

Subjects

Computer Science

Topics
preservation image processing machine learning algorithms privacy signal processing classification perturbations

MediaEval 2019: concealed FGSM perturbations for privacy preservation (2019)

Abstract This work tackles the Pixel Privacy task put forth by MediaEval 2019. Our goal is to decrease the accuracy of a classification algorithm while preserving the original image quality. We use the fast gradient sign method, which normally has a corrupting influence on image appeal, and devise two methods to minimize the damage. The first approach uses a map that is a combination of salient and flat areas. Perturbations are more noticeable in these locations, and so are directed away from them. The second approach adds the gradient of an aesthetic algorithm to the gradient of the attacking algorithm to guide the perturbations towards a direction that preserves appeal. We make our code available at: https://git.io/JesXr
Collections Ireland -> Dublin City University -> Publication Type = Conference or Workshop Item
Ireland -> Dublin City University -> Subject = Computer Science: Image processing
Ireland -> Dublin City University -> Status = Published
Ireland -> Dublin City University -> Subject = Engineering: Signal processing
Ireland -> Dublin City University -> DCU Faculties and Centres = DCU Faculties and Schools: Faculty of Engineering and Computing: School of Electronic Engineering
Ireland -> Dublin City University -> Subject = Computer Science: Algorithms
Ireland -> Dublin City University -> DCU Faculties and Centres = Research Initiatives and Centres: INSIGHT Centre for Data Analytics
Ireland -> Dublin City University -> Subject = Computer Science: Machine learning

Full list of authors on original publication

Kevin McGuinness, Suzanne Little, Panagiotis Linardos

Experts in our system

1
Kevin McGuinness
Dublin City University
Total Publications: 93
 
2
Suzanne Little
Dublin City University
Total Publications: 36