Summary of Cauchy-schwarz Divergence Information Bottleneck For Regression, by Shujian Yu et al.
Cauchy-Schwarz Divergence Information Bottleneck for Regression
by Shujian Yu, Xi Yu, Sigurd Løkse, Robert Jenssen, Jose C. Principe
First submitted to arxiv on: 27 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Information Theory (cs.IT); Machine Learning (stat.ML)
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 Cauchy-Schwarz Information Bottleneck (CS-IB) approach improves the generalization, robustness, and explainability of deep neural networks by finding a minimum sufficient representation. This is achieved by striking a trade-off between compression and prediction terms expressed in terms of mutual information. The CS-IB method moves away from mean squared error-based regression and avoids variational approximations or distributional assumptions. Experimental results demonstrate strong adversarial robustness guarantees and superior performance on six real-world regression tasks compared to other popular deep IB approaches. The code is available at https://github.com/SJYuCNEL/Cauchy-Schwarz-Information-Bottleneck. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper develops a new way to apply the Information Bottleneck (IB) principle to the regression problem using deep neural networks. It proposes a Cauchy-Schwarz (CS) version of IB that avoids mean squared error-based regression and variational approximations or distributional assumptions. The CS-IB method is shown to improve generalization, robustness, and explainability. Results are presented on six real-world regression tasks, demonstrating better performance than other popular deep IB approaches. |
Keywords
» Artificial intelligence » Generalization » Regression
