PENGUIN: Enhancing Transformer with Periodic-Nested Group Attention for Long-term Time Series Forecasting

Tian Sun, Yuqi Chen, Weiwei Sun
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:703-711, 2026.

Abstract

Despite advances in the Transformer architecture, their effectiveness for long-term time series forecasting (LTSF) remains controversial. In this paper, we investigate the potential of integrating explicit periodicity modeling into the self-attention mechanism to enhance the performance of Transformer-based architectures for LTSF. Specifically, we propose PENGUIN, a simple yet effective periodic-nested group attention mechanism. Our approach introduces a periodic-aware relative attention bias to directly capture periodic structures and a grouped multi-query attention mechanism to handle multiple coexisting periodicities (e.g., daily and weekly cycles) within time series data. Extensive experiments across diverse benchmarks demonstrate that PENGUIN consistently outperforms both MLP-based and Transformer-based models. Code is available at \url{https://github.com/ysygMhdxw/AISTATS2026_PENGUIN.}

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-sun26b, title = { PENGUIN: Enhancing Transformer with Periodic-Nested Group Attention for Long-term Time Series Forecasting }, author = {Sun, Tian and Chen, Yuqi and Sun, Weiwei}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {703--711}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/sun26b/sun26b.pdf}, url = {https://proceedings.mlr.press/v300/sun26b.html}, abstract = { Despite advances in the Transformer architecture, their effectiveness for long-term time series forecasting (LTSF) remains controversial. In this paper, we investigate the potential of integrating explicit periodicity modeling into the self-attention mechanism to enhance the performance of Transformer-based architectures for LTSF. Specifically, we propose PENGUIN, a simple yet effective periodic-nested group attention mechanism. Our approach introduces a periodic-aware relative attention bias to directly capture periodic structures and a grouped multi-query attention mechanism to handle multiple coexisting periodicities (e.g., daily and weekly cycles) within time series data. Extensive experiments across diverse benchmarks demonstrate that PENGUIN consistently outperforms both MLP-based and Transformer-based models. Code is available at \url{https://github.com/ysygMhdxw/AISTATS2026_PENGUIN.} } }
Endnote
%0 Conference Paper %T PENGUIN: Enhancing Transformer with Periodic-Nested Group Attention for Long-term Time Series Forecasting %A Tian Sun %A Yuqi Chen %A Weiwei Sun %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-sun26b %I PMLR %P 703--711 %U https://proceedings.mlr.press/v300/sun26b.html %V 300 %X Despite advances in the Transformer architecture, their effectiveness for long-term time series forecasting (LTSF) remains controversial. In this paper, we investigate the potential of integrating explicit periodicity modeling into the self-attention mechanism to enhance the performance of Transformer-based architectures for LTSF. Specifically, we propose PENGUIN, a simple yet effective periodic-nested group attention mechanism. Our approach introduces a periodic-aware relative attention bias to directly capture periodic structures and a grouped multi-query attention mechanism to handle multiple coexisting periodicities (e.g., daily and weekly cycles) within time series data. Extensive experiments across diverse benchmarks demonstrate that PENGUIN consistently outperforms both MLP-based and Transformer-based models. Code is available at \url{https://github.com/ysygMhdxw/AISTATS2026_PENGUIN.}
APA
Sun, T., Chen, Y. & Sun, W.. (2026). PENGUIN: Enhancing Transformer with Periodic-Nested Group Attention for Long-term Time Series Forecasting . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:703-711 Available from https://proceedings.mlr.press/v300/sun26b.html.

Related Material