Summary of How Much Can Time-related Features Enhance Time Series Forecasting?, by Chaolv Zeng et al.
How Much Can Time-related Features Enhance Time Series Forecasting?
by Chaolv Zeng, Yuan Tian, Guanjie Zheng, Yunjun Gao
First submitted to arxiv on: 2 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Machine Learning (stat.ML)
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 new approach to long-term time series forecasting (LTSF) by explicitly incorporating time-related features into the model. The current methods have been focused on capturing cross-time and cross-variate dependencies within historical data, but neglecting the importance of time-related encoding. The Time Stamp Forecaster (TimeSter) module is designed to address this gap and enhance the backbone’s forecasting performance. By integrating TimeSter with a linear backbone, the model, TimeLinear, achieves significant improvements in performance on benchmark datasets such as Electricity and Traffic, reducing Mean Squared Error (MSE) by an average of 23%. The proposed method also maintains exceptional computational efficiency, delivering results comparable to state-of-the-art models while using fewer parameters. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research paper is about improving the way computers forecast future events based on past data. Right now, most forecasting methods focus on finding patterns in the data, but they forget to include important details like time of day, month, or season. This can make it hard for models to capture trends that happen at specific times. The authors introduce a new module called Time Stamp Forecaster that helps fix this problem. By combining this module with a simple forecasting model, they get much better results on some famous benchmark datasets. This new approach is also very efficient and can give similar or better results than other top models while using fewer calculations. |
Keywords
» Artificial intelligence » Mse » Time series
