Summary of Evolving Multi-scale Normalization For Time Series Forecasting Under Distribution Shifts, by Dalin Qin et al.
Evolving Multi-Scale Normalization for Time Series Forecasting under Distribution Shifts
by Dalin Qin, Yehui Li, Weiqi Chen, Zhaoyang Zhu, Qingsong Wen, Liang Sun, Pierre Pinson, Yi Wang
First submitted to arxiv on: 29 Sep 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 The proposed EvoMSN framework tackles the distribution shift problem in long-term time series forecasting by introducing a novel, model-agnostic approach that adapts to shifting distributions. The framework combines flexible normalization and denormalization using multi-scale statistics prediction and adaptive ensembling. An evolving optimization strategy updates the forecasting model and statistics prediction module collaboratively to track distribution shifts. EvoMSN is evaluated on five mainstream forecasting methods, demonstrating its superiority over existing approaches. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new way to improve long-term time series forecasting by adapting to changes in data patterns. It uses a special kind of normalization that can adjust to changing conditions and update itself as it goes along. This helps the model make better predictions when the underlying data is shifting or changing over time. The approach is tested on several different methods and shown to be more accurate than other existing techniques. |
Keywords
» Artificial intelligence » Optimization » Time series
