EUVIMEDEUVIMED entdecken
Crossref

P‐254: Late‐News Poster: Dual‐dimension Graph Convolution Enhanced Transformer for Time Series Forecasting

Zheng Jin, Yuyu Liu, Xingqun Jiang

SID Symposium Digest of Technical Papers · 2026-06 · 57 (1) · 1891-1894

DOI: 10.1002/sdtp.70001_514 ↗

Kurz gefasst

Fachartikel in der Zeitschrift SID Symposium Digest of Technical Papers, online erschienen am 11.09.2026. Verfasst von Zheng Jin und 2 weiteren Autor:innen. Die Arbeit nennt 11 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Time series forecasting is fundamental in numerous fields, yet effectively modeling the complex temporal and variable dependencies within data remains a challenge. While Transformer has shown superiority by capturing long‐term dependencies via self‐attention, it entangles temporal and variable information within a single attention matrix, thus lacking an explicit mechanism for decoupled dependency modeling. To address this, we propose a novel Dual‐dimension Graph Convolution Enhanced Transformer (DGConvEnformer), which constructs two graph adjacency matrices from the attention scores to explicitly separate temporal and variable dependencies. These graphs are encoded by Graph Attention Network (GAT) and Graph Convolutional Network (GCN) modules for topological feature enhancement. Experimental results across multiple datasets demonstrate that DGConvEnformer achieves superior forecasting performance compared to various baselines, validating its capability of modeling complex dependencies in time series.

Zitieren

APA

Jin, Z., Liu, Y., & Jiang, X. (2026). P‐254: Late‐News Poster: Dual‐dimension Graph Convolution Enhanced Transformer for Time Series Forecasting. SID Symposium Digest of Technical Papers, 57(1), 1891-1894. https://doi.org/10.1002/sdtp.70001_514

Vancouver

Jin Z, Liu Y, Jiang X. P‐254: Late‐News Poster: Dual‐dimension Graph Convolution Enhanced Transformer for Time Series Forecasting. SID Symposium Digest of Technical Papers. 2026;57(1):1891-1894. doi:10.1002/sdtp.70001_514

BibTeX
@article{Jin2026,
  author = {Jin, Zheng and Liu, Yuyu and Jiang, Xingqun},
  title = {P‐254: Late‐News Poster: Dual‐dimension Graph Convolution Enhanced Transformer for Time Series Forecasting},
  journal = {SID Symposium Digest of Technical Papers},
  year = {2026},
  volume = {57},
  number = {1},
  pages = {1891--1894},
  doi = {10.1002/sdtp.70001_514}
}
Kostenlos starten →