Gridded Transformer Neural Processes for Spatio-Temporal Data

Matthew Ashman, Cristiana Diaconu, Eric Langezaal, Adrian Weller, Richard E Turner
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:1722-1761, 2025.

Abstract

Effective modelling of large-scale spatio-temporal datasets is essential for many domains, yet existing approaches often impose rigid constraints on the input data, such as requiring them to lie on fixed-resolution grids. With the rise of foundation models, the ability to process diverse, heterogeneous data structures is becoming increasingly important. Neural processes (NPs), particularly transformer neural processes (TNPs), offer a promising framework for such tasks, but struggle to scale to large spatio-temporal datasets due to the lack of an efficient attention mechanism. To address this, we introduce gridded pseudo-token TNPs which employ specialised encoders and decoders to handle unstructured data and utilise a processor comprising gridded pseudo-tokens with efficient attention mechanisms. Furthermore, we develop equivariant gridded TNPs for applications where exact or approximate translation equivariance is a useful inductive bias, improving accuracy and training efficiency. Our method consistently outperforms a range of strong baselines in various synthetic and real-world regression tasks involving large-scale data, while maintaining competitive computational efficiency. Experiments with weather data highlight the potential of gridded TNPs and serve as just one example of a domain where they can have a significant impact.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-ashman25a, title = {Gridded Transformer Neural Processes for Spatio-Temporal Data}, author = {Ashman, Matthew and Diaconu, Cristiana and Langezaal, Eric and Weller, Adrian and Turner, Richard E}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {1722--1761}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/ashman25a/ashman25a.pdf}, url = {https://proceedings.mlr.press/v267/ashman25a.html}, abstract = {Effective modelling of large-scale spatio-temporal datasets is essential for many domains, yet existing approaches often impose rigid constraints on the input data, such as requiring them to lie on fixed-resolution grids. With the rise of foundation models, the ability to process diverse, heterogeneous data structures is becoming increasingly important. Neural processes (NPs), particularly transformer neural processes (TNPs), offer a promising framework for such tasks, but struggle to scale to large spatio-temporal datasets due to the lack of an efficient attention mechanism. To address this, we introduce gridded pseudo-token TNPs which employ specialised encoders and decoders to handle unstructured data and utilise a processor comprising gridded pseudo-tokens with efficient attention mechanisms. Furthermore, we develop equivariant gridded TNPs for applications where exact or approximate translation equivariance is a useful inductive bias, improving accuracy and training efficiency. Our method consistently outperforms a range of strong baselines in various synthetic and real-world regression tasks involving large-scale data, while maintaining competitive computational efficiency. Experiments with weather data highlight the potential of gridded TNPs and serve as just one example of a domain where they can have a significant impact.} }
Endnote
%0 Conference Paper %T Gridded Transformer Neural Processes for Spatio-Temporal Data %A Matthew Ashman %A Cristiana Diaconu %A Eric Langezaal %A Adrian Weller %A Richard E Turner %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-ashman25a %I PMLR %P 1722--1761 %U https://proceedings.mlr.press/v267/ashman25a.html %V 267 %X Effective modelling of large-scale spatio-temporal datasets is essential for many domains, yet existing approaches often impose rigid constraints on the input data, such as requiring them to lie on fixed-resolution grids. With the rise of foundation models, the ability to process diverse, heterogeneous data structures is becoming increasingly important. Neural processes (NPs), particularly transformer neural processes (TNPs), offer a promising framework for such tasks, but struggle to scale to large spatio-temporal datasets due to the lack of an efficient attention mechanism. To address this, we introduce gridded pseudo-token TNPs which employ specialised encoders and decoders to handle unstructured data and utilise a processor comprising gridded pseudo-tokens with efficient attention mechanisms. Furthermore, we develop equivariant gridded TNPs for applications where exact or approximate translation equivariance is a useful inductive bias, improving accuracy and training efficiency. Our method consistently outperforms a range of strong baselines in various synthetic and real-world regression tasks involving large-scale data, while maintaining competitive computational efficiency. Experiments with weather data highlight the potential of gridded TNPs and serve as just one example of a domain where they can have a significant impact.
APA
Ashman, M., Diaconu, C., Langezaal, E., Weller, A. & Turner, R.E.. (2025). Gridded Transformer Neural Processes for Spatio-Temporal Data. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:1722-1761 Available from https://proceedings.mlr.press/v267/ashman25a.html.

Related Material