Summary of Knowledge-aware Neuron Interpretation For Scene Classification, by Yong Guan et al.
Knowledge-Aware Neuron Interpretation for Scene Classification
by Yong Guan, Freddy Lecue, Jiaoyan Chen, Ru Li, Jeff Z. Pan
First submitted to arxiv on: 29 Jan 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 Medium Difficulty summary: A novel knowledge-aware neuron interpretation framework is proposed to explain neural models’ predictions for image scene classification. Current methods lack concept completeness, fusion, and verification, which are addressed in this work. A core concept selection approach based on the ConceptNet knowledge graph gauges the completeness of concepts, providing better prediction explanations than baselines. Additionally, a concept filtering method integrates semantically-equivalent concepts, yielding over 23% point gain in neuron behavior interpretation. The framework also enables model manipulation, demonstrating that core concepts can improve the original model’s performance by over 26%. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Low Difficulty summary: A new way is developed to explain how artificial intelligence models make predictions about what they see in images. Currently, these explanations are not very good because they don’t consider all the important details or combine related ideas correctly. This new method uses a special database of concepts and connections to help AI models provide better explanations for their predictions. It also helps improve the AI model’s own performance by over 26%! |
