Summary of Towards Accurate and Robust Architectures Via Neural Architecture Search, by Yuwei Ou et al.
Towards Accurate and Robust Architectures via Neural Architecture Search
by Yuwei Ou, Yuqi Feng, Yanan Sun
First submitted to arxiv on: 9 May 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Cryptography and Security (cs.CR); 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 The proposed ARNAS framework searches for accurate and robust architectures for adversarial training. By designing an accurate and robust search space, the architecture can obtain both accuracy and robustness by deploying accurate and robust structures to their sensitive positions. A differentiable multi-objective search strategy is also proposed, which performs gradient descent towards directions that are beneficial for both natural loss and adversarial loss. Experimental results show that the searched architecture has strong robustness with competitive accuracy, and breaks the traditional idea that NAS-based architectures cannot transfer well to complex tasks in robustness scenarios. The framework can be applied to hand-crafting and automatically designing of accurate and robust architectures. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new way to design neural networks that are good at defending against cyber attacks. Instead of just trying to make the network more accurate, it also tries to make it more robust by using different parts of the network in different ways. This helps the network to be better at handling unexpected inputs and to be more resistant to attacks. The paper shows that this approach can lead to networks that are both very accurate and very robust, which is important for applications where security is a top priority. |
Keywords
* Artificial intelligence * Gradient descent
