Summary of Fast and Simple Explainability For Point Cloud Networks, by Meir Yossef Levi and Guy Gilboa
Fast and Simple Explainability for Point Cloud Networks
by Meir Yossef Levi, Guy Gilboa
First submitted to arxiv on: 12 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
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 This paper proposes a novel explainable AI (XAI) method for point cloud data, which computes pointwise importance with respect to a trained network’s downstream task. This approach allows for better understanding of the network properties, crucial for safety-critical applications. The method, called Feature Based Interpretability (FBI), computes features’ norm per point before the bottleneck layer and achieves significant speedups compared to current XAI methods. Additionally, FBI outperforms existing approaches in terms of classification explainability. The proposed method can be used for debugging, visualization, online feedback, reducing uncertainty, and increasing robustness. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper introduces a new way to understand how artificial intelligence (AI) works with 3D data. It helps us see what’s important about the AI’s decision-making process. This is especially important when we want to make sure the AI doesn’t make mistakes that could hurt people or cause problems. The method, called Feature Based Interpretability, is fast and can be used on big datasets or large-scale architectures. It works better than other methods for understanding why an AI makes certain decisions. |
Keywords
* Artificial intelligence * Classification
