Recent advances in computer vision and machine learning provide huge benefits in a variety of applications that benefit humanity greatly (e.g. Biomedical image analysis for early detection of diseases such as cancer).
But these technologies that have been exploited for serious forms of image tampering attacks. Examples include morphing face images to undermine reliability of traveller identification by automated border control
Systems, as well as Deep-Fake image/video generation for unethical exploitation by political rivalries. This seminar aims to establish that Persistent Homology provides a rich source for detecting such attacks.