Summary of Double Oracle Neural Architecture Search For Game Theoretic Deep Learning Models, by Aye Phyu Phyu Aung et al.
Double Oracle Neural Architecture Search for Game Theoretic Deep Learning Models
by Aye Phyu Phyu Aung, Xinrun Wang, Ruiyu Wang, Hau Chan, Bo An, Xiaoli Li, J. Senthilnath
First submitted to arxiv on: 7 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Science and Game Theory (cs.GT)
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 approach trains deep learning models using game theory concepts, including Generative Adversarial Networks (GANs) and Adversarial Training (AT). The method deploys a double-oracle framework, utilizing best response oracles for both GAN and AT. This framework extends the preliminary model DO-GAN to Adversarial Neural Architecture Search (NAS) for GAN and NAS for AT algorithms. The approach generalizes player strategies as trained models of generators and discriminators from best response oracles, computes meta-strategies using linear programming, and prunes weakly-dominated players’ strategies to ensure scalability. Experiments on MNIST, CIFAR-10, and TinyImageNet demonstrate significant improvements in both subjective qualitative evaluation and quantitative metrics compared to base architectures. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper uses game theory to train deep learning models. It’s like a competition between different teams, where each team tries to do better than the others. The approach is called DO-GAN, and it helps machines learn from each other. The researchers also tested this idea on real-world problems, like recognizing images, and showed that it works well. |
Keywords
» Artificial intelligence » Deep learning » Gan
