Conv-like Scale-Fusion Time Series Transformer: A Multi-Scale Representation for Variable-Length Long Time Series
Kai Zhang, Siming Sun, Zhengyu Fan, Qinmin Yang, Xuejun Jiang · 2025
Read Conv-like Scale-Fusion Time Series Transformer: A Multi-Scale Representation for Variable-Length Long Time Series by Kai Zhang, Siming Sun, Zhengyu Fan, Qinmin Yang, Xuejun Jiang (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-2509-17845.txt.
Description
Kai Zhang and 4 co-authors build a time-series model shaped like the pyramid form of classic image nets. Their 2025 paper deals with long records that vary in length, such as sensor or market data. Large attention models can waste traits and fail to travel to new data.
Their frame cuts time in patches, shrinks the time span step by step, and widens trait channels. A cross-scale step blends traits from fine and coarse views. A log-scale fix helps with records of mixed length. Tests show cleaner traits, less waste, and top scores on forecast and sort tasks.
Catalog record
| CREATOR | Kai Zhang, Siming Sun, Zhengyu Fan, Qinmin Yang, Xuejun Jiang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-17845