TwinFormer: A Dual-Level Transformer for Long-Sequence Time-Series Forecasting

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kumavat, Mahima, Maheshwari, Aditya
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911316711571456
author Kumavat, Mahima
Maheshwari, Aditya
author_facet Kumavat, Mahima
Maheshwari, Aditya
contents TwinFormer is a hierarchical Transformer for long-sequence time-series forecasting. It divides the input into non-overlapping temporal patches and processes them in two stages: (1) a Local Informer with top-$k$ Sparse Attention models intra-patch dynamics, followed by mean pooling; (2) a Global Informer captures long-range inter-patch dependencies using the same top-$k$ attention. A lightweight GRU aggregates the globally contextualized patch tokens for direct multi-horizon prediction. The resulting architecture achieves linear $O(kLd)$ time and memory complexity. On eight real-world benchmarking datasets from six different domains, including weather, stock price, temperature, power consumption, electricity, and disease, and forecasting horizons $96-720$, TwinFormer secures $27$ positions in the top two out of $34$. Out of the $27$, it achieves the best performance on MAE and RMSE at $17$ places and $10$ at the second-best place on MAE and RMSE. This consistently outperforms PatchTST, iTransformer, FEDformer, Informer, and vanilla Transformers. Ablations confirm the superiority of top-$k$ Sparse Attention over ProbSparse and the effectiveness of GRU-based aggregation. Code is available at this repository: https://github.com/Mahimakumavat1205/TwinFormer.
format Preprint
id arxiv_https___arxiv_org_abs_2512_12301
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TwinFormer: A Dual-Level Transformer for Long-Sequence Time-Series Forecasting
Kumavat, Mahima
Maheshwari, Aditya
Machine Learning
TwinFormer is a hierarchical Transformer for long-sequence time-series forecasting. It divides the input into non-overlapping temporal patches and processes them in two stages: (1) a Local Informer with top-$k$ Sparse Attention models intra-patch dynamics, followed by mean pooling; (2) a Global Informer captures long-range inter-patch dependencies using the same top-$k$ attention. A lightweight GRU aggregates the globally contextualized patch tokens for direct multi-horizon prediction. The resulting architecture achieves linear $O(kLd)$ time and memory complexity. On eight real-world benchmarking datasets from six different domains, including weather, stock price, temperature, power consumption, electricity, and disease, and forecasting horizons $96-720$, TwinFormer secures $27$ positions in the top two out of $34$. Out of the $27$, it achieves the best performance on MAE and RMSE at $17$ places and $10$ at the second-best place on MAE and RMSE. This consistently outperforms PatchTST, iTransformer, FEDformer, Informer, and vanilla Transformers. Ablations confirm the superiority of top-$k$ Sparse Attention over ProbSparse and the effectiveness of GRU-based aggregation. Code is available at this repository: https://github.com/Mahimakumavat1205/TwinFormer.
title TwinFormer: A Dual-Level Transformer for Long-Sequence Time-Series Forecasting
topic Machine Learning
url https://arxiv.org/abs/2512.12301