Summary of Scalable Transformer For High Dimensional Multivariate Time Series Forecasting, by Xin Zhou et al.
Scalable Transformer for High Dimensional Multivariate Time Series Forecasting
by Xin Zhou, Weiqing Wang, Wray Buntine, Shilin Qu, Abishek Sriramulu, Weicong Tan, Christoph Bergmeir
First submitted to arxiv on: 8 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
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 Scalable Transformer for High-Dimensional Multivariate Time Series Forecasting (STHD) tackles the limitations of existing channel-dependent models in handling high-dimensional MTS data. By analyzing the reasons behind their suboptimal performance, researchers identified two primary issues: introduced noise from unrelated series and challenges in training strategies due to high-dimensional data. To address these concerns, STHD incorporates three components: Relation Matrix Sparsity to limit noise and alleviate memory issues, ReIndex for flexible batch size settings and diverse training data, and a Transformer to handle 2-D inputs and capture channel dependencies. Experimental results on three high-dimensional datasets (Crime-Chicago, Wiki-People, and Traffic) demonstrate STHD’s considerable improvement. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The study focuses on improving deep models for multivariate time series forecasting. Researchers found that existing models struggle with high-dimensional data, leading to poor performance. To solve this problem, they created a new model called Scalable Transformer for High-Dimensional Multivariate Time Series Forecasting (STHD). This model has three parts: one helps remove noise from the data, another makes it easier to train, and the last part is a special kind of AI called a transformer that helps capture relationships between different types of data. The new model worked well on several large datasets. |
Keywords
* Artificial intelligence * Time series * Transformer
