PRAIRIE VIEW, Texas (January 13, 2023) — Dr. Yonghui Wang, assistant professor of computer science at Prairie View A&M University (PVAMU), was awarded $124,640 by the Princeton Alliance for Collaborative Research and Innovation (PACRI) program for his project “Deepfakes Detection: Solutions for a Growing Cybersecurity Concern.” The PACRI initiative provides funding for research projects co-led by principal investigators from Princeton University and a PACRI partner institution. (PVAMU is one of five institutions that partners with PACRI.)
Yonghui Wang, Ph.D., assistant professor of Computer Science
In collaboration with Princeton’s principal investigator (PI) Jia Deng, assistant professor of computer science, the project aims to address the rising and serious problem of deepfakes, which Wikipedia defines as “synthetic media in which a person in an existing image or video is replaced with someone else’s likeness.”
Numerous deepfakes circulating the internet involve celebrities, politicians, and other notable figures saying or doing things they never actually said or did. Famous examples include an entire TikTok account dedicated to deepfakes of Tom Cruise, and a video of former President Barack Obama making a public service announcement he never made. The advanced artificial intelligence technology is also being used by cybercriminals to target businesses, driving data breaches and financial scams. 
“In this interconnected age, deepfakes can be circulated quickly and target a massive audience, resulting in the spread of misinformation and posing a serious threat to personal privacy,” said PVAMU Vice President of Research & Innovation Magesh Rajan, Ph.D., P.E., MBA. “Dr. Wang’s project reinforces PVAMU’s commitment to exploring innovative solutions to concerns impacting society on a global scale.”
Given the advancement of artificial intelligence (AI), and especially deep learning technology – a form of AI and machine learning that teaches computers how to process data in a way that mimics the human brain – deepfake generative algorithms can now produce image, audio, and video representations realistic enough to fool any audience.
“Fake images and videos generated with the help of deep learning technology become highly realistic and extremely hard to detect,” said Dr. Wang. His collaborative project explores new methods for identifying deepfakes by developing “algorithms to detect falsified images and videos by examining inconsistencies in terms of high-level perceptual properties, such as lighting, defocus, optical flow, etc.,” said Dr. Wang.
Both graduate and undergraduate students will be trained in this project and the research results will be incorporated into Dr. Wang’s and Deng’s teachings. This is part of a global effort undertaken by major tech companies – such as Facebook, Amazon Web Services (AWS), and Microsoft – and researchers who are trying to design successful solutions to thwart these growing threats.
###
By Leigh Badrigian