Multiresolution Tensor Learning for Efficient and Interpretable Spatial Analysis

Jung Yeon Park, Kenneth Carr, Stephan Zheng, Yisong Yue, Rose Yu
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:7499-7509, 2020.

Abstract

Efficient and interpretable spatial analysis is crucial in many fields such as geology, sports, and climate science. Tensor latent factor models can describe higher-order correlations for spatial data. However, they are computationally expensive to train and are sensitive to initialization, leading to spatially incoherent, uninterpretable results. We develop a novel Multiresolution Tensor Learning (MRTL) algorithm for efficiently learning interpretable spatial patterns. MRTL initializes the latent factors from an approximate full-rank tensor model for improved interpretability and progressively learns from a coarse resolution to the fine resolution to reduce computation. We also prove the theoretical convergence and computational complexity of MRTL. When applied to two real-world datasets, MRTL demonstrates 4 5x speedup compared to a fixed resolution approach while yielding accurate and interpretable latent factors.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-park20a, title = {Multiresolution Tensor Learning for Efficient and Interpretable Spatial Analysis}, author = {Park, Jung Yeon and Carr, Kenneth and Zheng, Stephan and Yue, Yisong and Yu, Rose}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {7499--7509}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/park20a/park20a.pdf}, url = {https://proceedings.mlr.press/v119/park20a.html}, abstract = {Efficient and interpretable spatial analysis is crucial in many fields such as geology, sports, and climate science. Tensor latent factor models can describe higher-order correlations for spatial data. However, they are computationally expensive to train and are sensitive to initialization, leading to spatially incoherent, uninterpretable results. We develop a novel Multiresolution Tensor Learning (MRTL) algorithm for efficiently learning interpretable spatial patterns. MRTL initializes the latent factors from an approximate full-rank tensor model for improved interpretability and progressively learns from a coarse resolution to the fine resolution to reduce computation. We also prove the theoretical convergence and computational complexity of MRTL. When applied to two real-world datasets, MRTL demonstrates 4 5x speedup compared to a fixed resolution approach while yielding accurate and interpretable latent factors.} }
Endnote
%0 Conference Paper %T Multiresolution Tensor Learning for Efficient and Interpretable Spatial Analysis %A Jung Yeon Park %A Kenneth Carr %A Stephan Zheng %A Yisong Yue %A Rose Yu %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-park20a %I PMLR %P 7499--7509 %U https://proceedings.mlr.press/v119/park20a.html %V 119 %X Efficient and interpretable spatial analysis is crucial in many fields such as geology, sports, and climate science. Tensor latent factor models can describe higher-order correlations for spatial data. However, they are computationally expensive to train and are sensitive to initialization, leading to spatially incoherent, uninterpretable results. We develop a novel Multiresolution Tensor Learning (MRTL) algorithm for efficiently learning interpretable spatial patterns. MRTL initializes the latent factors from an approximate full-rank tensor model for improved interpretability and progressively learns from a coarse resolution to the fine resolution to reduce computation. We also prove the theoretical convergence and computational complexity of MRTL. When applied to two real-world datasets, MRTL demonstrates 4 5x speedup compared to a fixed resolution approach while yielding accurate and interpretable latent factors.
APA
Park, J.Y., Carr, K., Zheng, S., Yue, Y. & Yu, R.. (2020). Multiresolution Tensor Learning for Efficient and Interpretable Spatial Analysis. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:7499-7509 Available from https://proceedings.mlr.press/v119/park20a.html.

Related Material