Summary of Adaptive Batch Normalization Networks For Adversarial Robustness, by Shao-yuan Lo et al.
Adaptive Batch Normalization Networks for Adversarial Robustness
by Shao-Yuan Lo, Vishal M. Patel
First submitted to arxiv on: 20 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
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 explores a new approach to defending deep networks against adversarial examples without relying on Adversarial Training (AT). The proposed Adaptive Batch Normalization Network (ABNN) uses adaptive Batch Normalization (BN) statistics generated by a pre-trained substitute model. ABNN is trained exclusively on clean data and learns to align the substitute model’s BN statistics, achieving improved robustness against strong attacks on image and video datasets. Additionally, ABNN outperforms AT-based approaches in terms of clean data performance and training time complexity. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In simple terms, this paper develops a new way to protect deep networks from fake images or videos that are designed to trick them. The method uses statistics generated by a similar network that was trained on real data, allowing the target network to learn how to recognize patterns in the same way. This approach is faster and more efficient than traditional methods and can be applied to various types of data. |
Keywords
» Artificial intelligence » Batch normalization
