Summary of Affakt: a Hierarchical Optimal Transport Based Method For Affective Facial Knowledge Transfer in Video Deception Detection, by Zihan Ji et al.
AFFAKT: A Hierarchical Optimal Transport based Method for Affective Facial Knowledge Transfer in Video Deception Detection
by Zihan Ji, Xuetao Tian, Ye Liu
First submitted to arxiv on: 12 Dec 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel method called AFFAKT is proposed to enhance the classification performance in video deception detection by transferring knowledge from a large facial expression dataset. The approach addresses two key challenges: determining how much knowledge to transfer and leveraging transferred knowledge effectively during inference. The H-OTKT module quantifies the optimal relation mapping between facial expression classes and deception samples, while the SRKB module retains invariant correlations through momentum updating. A sample-specific re-weighting strategy is used to fine-tune transferred knowledge. Experimental results on two datasets demonstrate superior performance, with interpretability studies revealing high associations between deception and negative affections. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to help machines detect fake videos is being developed by using facial expressions. The idea is to take information from a big dataset of people’s faces showing different emotions and use that knowledge to improve the accuracy of detecting when someone is lying or not telling the truth in a video. This method, called AFFAKT, helps solve two tricky problems: knowing how much to transfer this emotional knowledge and using it effectively. The results show that AFFAKT works better than other methods on two sets of data, and it also reveals some surprising insights about the connections between emotions and deception. |
Keywords
» Artificial intelligence » Classification » Inference
