Benchmarking Graph Representations and Graph Neural Networks for Multivariate Time Series Classification
Wennuo Yang, Shiling Wu, Yuzhi Zhou, Cheng Luo, Xilin He, Weicheng Xie, Linlin Shen, Siyang Song · 2025
Read Benchmarking Graph Representations and Graph Neural Networks for Multivariate Time Series Classification by Wennuo Yang, Shiling Wu, Yuzhi Zhou, Cheng Luo, Xilin He, Weicheng Xie, Linlin Shen, Siyang Song (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-08305.txt.
Description
Yang and seven co-authors present a benchmark study from twenty twenty-five on graph methods for multivariate time series sorting. Such series track many linked channels over time, as in health or finance data. Graph methods can expose links between channels that flat methods miss. No fair benchmark spanned the main design choices, though. The team build sixty variants from three node trait plans, four edge learning plans, and five network designs. All variants share one data pipeline and training scheme. They run on twenty-six public datasets.
Node trait choices sway scores most in their results. Adaptive edge learning beats rival edge plans, and charts show why. The code is posted in public for reuse. The authors aim to settle design guidance for the field.
Catalog record
| CREATOR | Wennuo Yang, Shiling Wu, Yuzhi Zhou, Cheng Luo, Xilin He, Weicheng Xie, Linlin Shen, Siyang Song |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-08305