Summary of Deep Frequency Derivative Learning For Non-stationary Time Series Forecasting, by Wei Fan et al.
Deep Frequency Derivative Learning for Non-stationary Time Series Forecasting
by Wei Fan, Kun Yi, Hangting Ye, Zhiyuan Ning, Qi Zhang, Ning An
First submitted to arxiv on: 29 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 This paper proposes a novel approach to address the distribution shift issue in time series forecasting by utilizing the whole frequency spectrum. The existing solutions manipulate statistical measures like mean and standard deviation, but this can be seen as transforming towards zero frequency component, which hinders forecasting performance. The proposed framework, DERITS, uses Frequency Derivative Transformation (FDT) to make signals derived from the frequency domain more stationary. Then, it applies Order-adaptive Fourier Convolution Network for adaptive frequency filtering and learning. Finally, a parallel-stacked architecture is used for multi-order derivation and fusion for forecasting. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us better predict what will happen in the future by using all of the information available in time series data. Right now, when we try to forecast what will happen next, we’re limited by how well our models understand the data’s distribution. To fix this, researchers have tried manipulating statistical measures like mean and standard deviation. However, this approach has its limitations. In this paper, scientists propose a new way to transform time series data using all of the information available in the frequency spectrum. This allows for better forecasting performance and more accurate predictions. |
Keywords
* Artificial intelligence * Time series
