Summary of From Movements to Metrics: Evaluating Explainable Ai Methods in Skeleton-based Human Activity Recognition, by Kimji N. Pellano et al.
From Movements to Metrics: Evaluating Explainable AI Methods in Skeleton-Based Human Activity Recognition
by Kimji N. Pellano, Inga Strümke, Espen Alexander F. Ihlen
First submitted to arxiv on: 20 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 paper bridges a gap in human activity recognition (HAR) using 3D skeleton data by evaluating existing eXplainable Artificial Intelligence (XAI) metrics, specifically faithfulness and stability, on Class Activation Mapping (CAM) and Gradient-weighted Class Activation Mapping (Grad-CAM). The study introduces a perturbation method respecting human biomechanical constraints to ensure realistic variations in human movement. The findings indicate that faithfulness may not be reliable for certain models like EfficientGCN, while stability is more dependable with slight input data perturbations. CAM and Grad-CAM produce similar explanations, leading to comparable XAI metric performance, highlighting the need for diversified metrics and new approaches. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us understand human activities better by testing how well we can explain what machines are thinking when they recognize our movements. The researchers looked at two special ways to make AI models more transparent: faithfulness and stability. They also developed a way to change the input data in a way that still looks like normal human movement. Their results show that faithfulness isn’t always reliable, but stability is a better measure. Surprisingly, two popular methods for explaining AI decisions produce very similar results. |
Keywords
* Artificial intelligence * Activity recognition
