Listen Watch and Read: Multimodal Information Authenticity Detection and Source Tracking

Our project aims to identify and detect falsified video content using multi-modal information and deep learning. Our team, comprising experts in speech, image, and text analysis, collaborates closely, sharing knowledge to develop innovative machine learning architectures. Additionally, a criminal science expert provides forensic insights, enhancing our efforts to create a robust counterfeit content detection system. We are confident in our ability to develop a sophisticated solution to counteract illegal uses of multimedia generation technologies.

Closely working with the Taiwan FactCheck Center. Proposed an interactive game-based claim labeling system and collected claim detection dataset. Developed an claim verification system that can detect fake news and misinformation.