Summary of Xpatch: Dual-stream Time Series Forecasting with Exponential Seasonal-trend Decomposition, by Artyom Stitsyuk and Jaesik Choi
xPatch: Dual-Stream Time Series Forecasting with Exponential Seasonal-Trend Decomposition
by Artyom Stitsyuk, Jaesik Choi
First submitted to arxiv on: 23 Dec 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 The novel dual-stream architecture, xPatch, designed for time-series forecasting, leverages exponential decomposition and patching techniques to improve performance. By introducing a seasonal-trend exponential decomposition module and combining it with an MLP-based linear stream and a CNN-based non-linear stream, xPatch outperforms traditional transformer-based models in exploiting temporal relations within time series data. Additionally, the proposed arctangent loss function and sigmoid learning rate adjustment scheme prevent overfitting and enhance forecasting accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary xPatch is a new way to forecast future events based on past patterns. It’s like a special tool that helps us predict what will happen next in a series of events. The team created xPatch by combining two different types of models: one that works well with straight lines and another that excels at recognizing patterns. This allows xPatch to capture both the big picture trends and the smaller, more detailed fluctuations in the data. By using this new approach, we can make better predictions about what will happen next. |
Keywords
» Artificial intelligence » Cnn » Loss function » Overfitting » Sigmoid » Time series » Transformer
