UM students win global deepfake detection contest

2024-09-20 02:56
BY Rui Pastorin
Comment:0

The University of Macau (UM) said in a statement yesterday that its team from the Faculty of Science and Technology (FST) won the championship of the Global Multimedia Deepfake Detection Challenge held during the 2024 Inclusion Conference on the Bund in Shanghai, taking home a 100,000-yuan prize.

The recent contest, hosted on the data science competition platform Kaggle, had a special focus on practical AI face-swapping exercises, aiming to help develop solutions to counter deepfake attacks in various scenarios, among others.

The team, trained by Department of Computer and Information Science Head Zhou Jiantao, comprised students Chen Yiming, Li Fengpeng, Li Kemou, Wong Ka Him, Li Zheng, Chen Xiangyu, Song Binbin, Xu Shuning, and Liu Jun, as well as UM alumnus Wu Haiwei, the statement noted. They competed against nearly 10,000 participants in over 7,000 teams from institutions and companies, such as Stanford University and ByteDance, from more than 20 countries and regions.

According to the statement, deepfake technology uses “advanced deep learning algorithms to create highly realistic facial animations and voice synthesis effects by learning from large amounts of video and image data”, aka “AI face-swapping” technology. The technology makes it difficult to determine video authenticity and increases risks of fraud, pornography, and other illegal activities.

Deepfakes, Wikipedia notes, have also gained attention for their potential use in creating child sexual abuse material, purported celebrity pornographic videos, revenge porn, fake news, hoaxes, bullying, and financial fraud.

The team’s winning solution introduced a “data type-aware unsupervised clustering method to reclassify datasets for simulating cross-domain testing scenarios” and “developed feature-level adversarial training algorithms and type augmentation algorithms”, enhancing data types and guided detection models to learn high-generalisation deepfake features adaptable to multiple types and scenarios.

Research findings are available on: https://github.com/HighwayWu/DeepFakeDefenders

This undated handout photo provided by the University of Macau (UM) yesterday shows a group photo of members of the winning team and the head of the Department of Computer and Information Science, Zhou Jiantao (fourth from right).


0 COMMENTS

Leave a Reply