Communications of the ACM
Filter Protects Against Deepfake Photos, Videos
A new algorithm adds a filter to videos and photos to prevent deepfakes, researchers say.
Credit: Lightspring/Shutterstock.com
Computer scientists at Boston University (BU) have created an algorithm that prevents deepfakes by adding an imperceptible filter to videos and photos before they are uploaded to the Internet.
If a deep neural network attempts to alter a protected image or video, the filter leaves the media unchanged, or completely distorted and rendered unusable as a deepfake.
The open source code has been made publicly available on GitHub.
Said BU's Sarah Adel Bargal, "We covered what we call 'white-box' attacks in our work, where the network and its parameters are known to the disruptor. A very important next step is to develop methods for 'black-box' attacks that can disrupt deepfake networks [in ways] inaccessible to the disruptor."
From NextGov.com
View Full Article
Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA
No entries found