Summary of Emotion Loss Attacking: Adversarial Attack Perception For Skeleton Based on Multi-dimensional Features, by Feng Liu et al.
Emotion Loss Attacking: Adversarial Attack Perception for Skeleton based on Multi-dimensional Features
by Feng Liu, Qing Xu, Qijian Zheng
First submitted to arxiv on: 28 Jun 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 The proposed adversarial attack method targets action recognizers for skeletal motions, addressing limitations in existing research by considering both dynamic and emotional features. The novel approach includes a dynamic distance function for measuring differences between skeleton graph sequences and introduces emotional features as complementary information. The constrained optimization problem is solved using Alternating Direction Method of Multipliers (ADMM), generating adversarial samples with improved imperceptibility to deceive classifiers. Experimental results demonstrate the method’s effectiveness on multiple action classifiers and datasets, with perturbations generated by this method showing better imperceptibility than others at the same l-norm magnitude. This work also provides a new idea for measuring distance between skeletal motions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers has developed a new way to trick computer systems that recognize human actions from videos or other data. Currently, these systems are easily fooled by fake data that is slightly different from real data. The new method creates fake data that is much harder to distinguish from real data, making it more difficult for the system to accurately recognize actions. This technique uses a combination of two types of features: dynamic features, which describe how the body moves over time, and emotional features, which capture the feeling or tone behind an action. The new method is able to create fake data that is much harder to distinguish from real data than previous methods, making it more difficult for action recognition systems to accurately recognize actions. |
Keywords
» Artificial intelligence » Optimization
