Summary of Explaining the Unexplained: Revealing Hidden Correlations For Better Interpretability, by Wen-dong Jiang et al.
Explaining the Unexplained: Revealing Hidden Correlations for Better Interpretability
by Wen-Dong Jiang, Chih-Yung Chang, Show-Jane Yen, Diptendu Sinha Roy
First submitted to arxiv on: 2 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- 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 This paper proposes RealExp, an interpretable machine learning method that decouples the Shapley Value into feature importance and correlation importance. Unlike existing methods, RealExp considers feature correlations and provides nuanced explanations by quantifying both individual feature contributions and their interactions. Additionally, the authors introduce a novel interpretability evaluation criterion focused on decision paths, going beyond accuracy-based metrics. Experimental results demonstrate that RealExp outperforms existing methods in interpretability on image classification and text sentiment analysis tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper introduces a new way to understand how deep learning models work. Right now, these models can be like “black boxes” – we don’t know why they make certain decisions. To fix this, the authors created RealExp, a method that breaks down what each feature in the data is doing and how it’s related to other features. This helps us understand not just individual features, but also how they work together. The paper also proposes a new way to measure how well an explanation works, going beyond just saying “this model is correct or not”. The results show that RealExp does better than existing methods at explaining why deep learning models make certain decisions. |
Keywords
» Artificial intelligence » Deep learning » Image classification » Machine learning
