FB pixel

Samsung researchers develop method of creating deepfakes from minimal source data

Categories Biometric R&D  |  Biometrics News
 

Samsung has developed an artificial intelligence system for creating deepfake video from as little source data as one image, CNET reports.

The Russia-based team of Samsung researchers published a paper and presented a video on “Few-Shot Adversarial Learning of Realistic Neural Talking Head models,” which shows the creation of “realistic neural talking heads” using neural networks. The researchers say it could be applied to videoconferencing, video games, and the special effects industry.

Dartmouth media forensics researcher Hany Farid calls the technique “another step in the evolution” of deepfakes. “Following the trend of the past year, this and related techniques require less and less data and are generating more and more sophisticated and compelling content,” he says.

A manipulated video of U.S. House Speaker Nancy Pelosi appearing to slur her speech recently proliferated around the internet.

The system begins with an extensive “meta-learning stage” in which it trains on the movement of faces in large quantities of video, and then applies it to a single image or handful of pictures. The final result shows imperfect details, particularly with a single image, and University of Albany Computer Science Professor Siwei Lyu says it tends to resemble whoever the face movement was modeled on. Lyu says it could save training time and make the model more generalizable, however.

China recently moved to ban deepfakes, and the AI Foundation and the Technical University of Munich have partnered to advance research on the problem.

Article Topics

 |   |   |   |   | 

Latest Biometrics News

 

Securing user trust and safeguarding platforms with biometric identity verification

Verified trust is the new currency: so says a new report from reusable verified identity and screening company Trua, looking…

 

Essex Police reveal impressive accuracy of LFR from Corsight, Digital Barriers

England’s Essex Police have performed 383,356 match attempts with live facial recognition software from Corsight AI and Digital Barriers, with…

 

US and UK refusal to sign Paris declaration shows divergence in AI strategy

The U.S. and the UK have declined to sign the Paris AI summit declaration, which seeks to establish a “human…

 

DHS’s compliance with AI privacy, civil liberties requirements lacking, IG says

The Department of Homeland Security (DHS) has made strides in developing policies and frameworks to govern its AI use, including…

 

Precise Biometrics: quarterlies, annuals, SEC actions

Feb 13, 2025 – Net sales for Precise Biometrics rose 15.7 percent percent from 75.1 million Swedish kronor (approximately US$7 million)…

 

YouTube, Meta lean into age assurance in 2025

In the past twelve months, age assurance for online content – a method for knowing that a user is of…

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Viewed This Week

Featured Company

Biometrics Insight, Opinion

Digital ID In-Depth

Biometrics White Papers

Biometrics Events