Summary of Deeplink-t: Deep Learning Inference For Time Series Data Using Knockoffs and Lstm, by Wenxuan Zuo et al.
DeepLINK-T: deep learning inference for time series data using knockoffs and LSTM
by Wenxuan Zuo, Zifan Zhu, Yuxuan Du, Yi-Chun Yeh, Jed A. Fuhrman, Jinchi Lv, Yingying Fan, Fengzhu Sun
First submitted to arxiv on: 5 Apr 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
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 introduces a novel method called Deep Learning Inference using Knockoffs for Time series data (DeepLINK-T) that combines deep learning with knockoff inference to control the false discovery rate (FDR) at a predetermined level. This method is designed to select significant time series variables in regression models while accommodating a wide variety of feature distributions and addressing dependencies across time and features. The approach involves an LSTM autoencoder for generating time series knockoff variables, an LSTM prediction network using both original and knockoff variables, and the application of the knockoffs framework for variable selection with FDR control. Simulation studies demonstrate the method’s capability to control FDR effectively while showcasing superior feature selection power compared to its non-time series counterpart. The paper also applies DeepLINK-T to three metagenomic data sets, validating its practical utility and effectiveness. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about a new way to analyze high-dimensional time series data, which is important for many real-world applications. It’s like trying to find the most important features in a big dataset that changes over time. The method uses deep learning and a technique called knockoff inference to make sure it doesn’t pick too many “important” features by chance. This approach works well with datasets that have different patterns at different times, which is common in many fields like medicine or environmental science. The researchers tested their method on some sample data and found that it worked really well. |
Keywords
* Artificial intelligence * Autoencoder * Deep learning * Feature selection * Inference * Lstm * Regression * Time series
