Summary of Batch-in-batch: a New Adversarial Training Framework For Initial Perturbation and Sample Selection, by Yinting Wu (1) et al.
Batch-in-Batch: a new adversarial training framework for initial perturbation and sample selection
by Yinting Wu, Pai Peng, Bo Cai, Le Li
First submitted to arxiv on: 6 Jun 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 The paper proposes a simple yet effective training framework called Batch-in-Batch (BB) to enhance models’ robustness against adversarial attacks. The BB framework involves generating multiple sets of perturbations from the original batch set and selecting samples strategically to avoid overconfident outputs. Experimental results on three benchmark datasets (CIFAR-10, SVHN, CIFAR-100) with two networks (PreActResNet18 and WideResNet28-10) show that models trained within the BB framework consistently outperform baseline models in adversarial accuracy, achieving a 13% improvement on the SVHN dataset. The paper also validates the efficiency of the proposed initial perturbation method and sample selection strategies. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper tries to make AI models more robust against fake or misleading information. They come up with a new way to train these models that involves creating many different versions of the same information, then picking the best ones. This helps the model learn to be less confident in its answers and avoid getting tricked by fake data. The authors tested their method on several datasets and found that it worked really well, especially when the fake data was very convincing. |
