Streaming Gaussian Dirichlet Random Fields for Spatial Predictions of High Dimensional Categorical Observations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Soucie, J. E. San, Sosik, H. M., Girdhar, Y.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929254724272128
author Soucie, J. E. San
Sosik, H. M.
Girdhar, Y.
author_facet Soucie, J. E. San
Sosik, H. M.
Girdhar, Y.
contents We present the Streaming Gaussian Dirichlet Random Field (S-GDRF) model, a novel approach for modeling a stream of spatiotemporally distributed, sparse, high-dimensional categorical observations. The proposed approach efficiently learns global and local patterns in spatiotemporal data, allowing for fast inference and querying with a bounded time complexity. Using a high-resolution data series of plankton images classified with a neural network, we demonstrate the ability of the approach to make more accurate predictions compared to a Variational Gaussian Process (VGP), and to learn a predictive distribution of observations from streaming categorical data. S-GDRFs open the door to enabling efficient informative path planning over high-dimensional categorical observations, which until now has not been feasible.
format Preprint
id arxiv_https___arxiv_org_abs_2402_15359
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Streaming Gaussian Dirichlet Random Fields for Spatial Predictions of High Dimensional Categorical Observations
Soucie, J. E. San
Sosik, H. M.
Girdhar, Y.
Robotics
Machine Learning
We present the Streaming Gaussian Dirichlet Random Field (S-GDRF) model, a novel approach for modeling a stream of spatiotemporally distributed, sparse, high-dimensional categorical observations. The proposed approach efficiently learns global and local patterns in spatiotemporal data, allowing for fast inference and querying with a bounded time complexity. Using a high-resolution data series of plankton images classified with a neural network, we demonstrate the ability of the approach to make more accurate predictions compared to a Variational Gaussian Process (VGP), and to learn a predictive distribution of observations from streaming categorical data. S-GDRFs open the door to enabling efficient informative path planning over high-dimensional categorical observations, which until now has not been feasible.
title Streaming Gaussian Dirichlet Random Fields for Spatial Predictions of High Dimensional Categorical Observations
topic Robotics
Machine Learning
url https://arxiv.org/abs/2402.15359