Summary of Deepdrk: Deep Dependency Regularized Knockoff For Feature Selection, by Hongyu Shen and Yici Yan and Zhizhen Zhao
DeepDRK: Deep Dependency Regularized Knockoff for Feature Selection
by Hongyu Shen, Yici Yan, Zhizhen Zhao
First submitted to arxiv on: 27 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 Model-X knockoff has gained attention for its guarantees in controlling the false discovery rate (FDR), but current implementations of the deep Model-X knockoff framework face limitations. The “swap property” that knockoffs require often struggles at the sample level, reducing selection power. To address these issues, we introduce Deep Dependency Regularized Knockoff (DeepDRK), a distribution-free deep learning method balancing FDR and power. We reformulate the knockoff model as a learning problem under multi-source adversarial attacks and employ an innovative perturbation technique to achieve lower FDR and higher power. Our model outperforms existing benchmarks across synthetic, semi-synthetic, and real-world datasets, particularly when sample sizes are small and data distributions are non-Gaussian. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Model-X knockoff has been popular for its ability to control the false discovery rate (FDR). But there have been problems with how it works. To fix this, we created a new way to do Model-X knockoff using deep learning that doesn’t rely on specific data distributions. Our method is called Deep Dependency Regularized Knockoff (DeepDRK). We made some changes to the model so it can handle different types of data and make better choices about what’s important. This worked well across many datasets, especially when there wasn’t much data or the data was very different from what we expected. |
Keywords
* Artificial intelligence * Attention * Deep learning
