With the rapid development of information dissemination promoted by social networks, massive multimedia information is shared on different social platforms through Internet. Despite its effectiveness in enriching our lives, such information brings serious security issues to the society, such as problems of fake news, privacy disclosure, illegal theft, and malicious tampering. At present, the study of technologies for multimedia forensics and security over social networks is still in its infancy, and the main challenges include the interference of social networks, the poor generalization ability, and the low security reliability. To address above difficulties, this project took the lead in establishing the analysis mechanism for different social platforms, developing highly robust forensic technologies against social multimedia, designing security analysis frameworks for forensic technologies, and robust feature learning algorithms with data privacy protection. Results of this project have been widely cited and applied by many scholars, and have also been reported by CCTV news, Macau government, Macau news TV and Weibo. Achievements of this project let Macau to keep an international leading position in this important research field.

Fig 1 The image forgery detection system “Fake Pictures Crusher” developed by the team in cooperation with Alibaba

Fig 2 Trustworthy deep model inference architecture for multimedia data