CaltechAUTHORS
  A Caltech Library Service

Multiresolution Tensor Learning for Efficient and Interpretable Spatial Analysis

Park, Jung Yeon and Carr, Kenneth Theo and Zhang, Stephan and Yue, Yisong and Yu, Rose (2020) Multiresolution Tensor Learning for Efficient and Interpretable Spatial Analysis. . (Unpublished) https://resolver.caltech.edu/CaltechAUTHORS:20200214-105610460

[img] PDF - Submitted Version
See Usage Policy.

1334Kb

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20200214-105610460

Abstract

Efficient and interpretable spatial analysis is crucial in many fields such as geology, sports, and climate science. Large-scale spatial data often contains complex higher-order correlations across features and locations. While tensor latent factor models can describe higher-order correlations, they are inherently computationally expensive to train. Furthermore, for spatial analysis, these models should not only be predictive but also be spatially coherent. However, latent factor models are sensitive to initialization and can yield inexplicable results. We develop a novel Multi-resolution 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 for an enormous computation speedup. We also prove the theoretical convergence and computational complexity of MRTL. When applied to two real-world datasets, MRTL demonstrates 4 ~ 5 times speedup compared to a fixed resolution while yielding accurate and interpretable models.


Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription
http://arxiv.org/abs/2002.05578arXivDiscussion Paper
ORCID:
AuthorORCID
Yue, Yisong0000-0001-9127-1989
Record Number:CaltechAUTHORS:20200214-105610460
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20200214-105610460
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:101304
Collection:CaltechAUTHORS
Deposited By: George Porter
Deposited On:14 Feb 2020 20:50
Last Modified:14 Feb 2020 20:50

Repository Staff Only: item control page