Summary of Bridging the Gap: Rademacher Complexity in Robust and Standard Generalization, by Jiancong Xiao et al.
Bridging the Gap: Rademacher Complexity in Robust and Standard Generalization
by Jiancong Xiao, Ruoyu Sun, Qi Long, Weijie J. Su
First submitted to arxiv on: 8 Jun 2024
Categories
- Main: Machine Learning (stat.ML)
- 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 A medium-difficulty summary: This paper investigates the issue of adversarially robust generalization in deep neural networks (DNNs), where training with adversarial examples leads to poor performance on test-time adversarial data. The authors build upon previous studies and propose new bounds for DNNs, showing that existing works often rely on surrogate losses or yield looser bounds compared to standard counterparts. The proposed bounds have a higher dependency on the width of the DNNs (m) or the dimension of the data (d), with an extra factor of at least O(sqrt(m)) or O(sqrt(d)). The authors’ work provides new insights into the challenge of achieving satisfactory robust generalization in DNNs, highlighting the importance of developing effective methods for training robust models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A low-difficulty summary: This paper explores a problem with deep learning, where training with fake data makes it hard to get good results on real data. The authors want to find better ways to train these “deep neural networks” so they can work well even when the test data is different from the training data. They look at how previous research has tried to solve this problem and propose new ideas for making deep learning models more robust. |
Keywords
» Artificial intelligence » Deep learning » Generalization
