Summary of Phase-driven Domain Generalizable Learning For Nonstationary Time Series, by Payal Mohapatra et al.
Phase-driven Domain Generalizable Learning for Nonstationary Time Series
by Payal Mohapatra, Lixu Wang, Qi Zhu
First submitted to arxiv on: 5 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Signal Processing (eess.SP)
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 PhASER framework is a novel time-series learning approach that effectively handles non-stationary data by incorporating phase information and leveraging residual connections for regularization. By augmenting phase, separating feature encoding into magnitude and phase modalities, and broadcasting phase with residual connections, PhASER outperforms state-of-the-art baselines on five benchmark datasets from human activity recognition, sleep-stage classification, and gesture recognition. The average improvement is 5% and up to 13%, demonstrating the framework’s ability to boost generalization in time-series classification tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary PhASER is a new way of learning patterns in continuous data that changes over time. This is important for many real-world applications like recognizing human activities, sleep stages, or gestures. The problem is that this kind of data is often unpredictable and varies greatly. PhASER solves this by looking at the “phase” of the data, which is a new way to understand patterns. It also separates the data into two parts: magnitude (amount) and phase (direction). This helps machines learn more effectively from changing data. The results show that PhASER works better than other methods on several datasets. |
Keywords
* Artificial intelligence * Activity recognition * Classification * Generalization * Gesture recognition * Regularization * Time series
