Gaussian Process Regression for Surface Interpolation скачать в хорошем качестве

Gaussian Process Regression for Surface Interpolation

2022.07.06, Zhiqiao Dong & Manan Mehta, University of Illinois Urbana-Champaign Filtered Kriging Lab tool can be found at: https://nanohub.org/tools/fkriging Part of Hands-on Data Science and Machine Learning Training Series at: https://nanohub.org/groups/ml/handson... Table of contents below. Gaussian process regression (GPR) is a nonparametric regression method with widespread applications in various scientific and engineering fields. In manufacturing, it has been used for surface interpolation that generates high-resolution surface estimations from coarser measurement data. This tutorial will introduce the fundamentals of GPR and its application to surface interpolation. We will also introduce a new technique called filtered kriging (FK), which uses a pre-filter to improve interpolation performance. The FK method will be illustrated using periodic surfaces manufactured by two photon lithography. Table of Contents: 00:00 Gaussian Process Regression for Surface Interpolation 00:53 A Motivating Example from Nanomanufacturing 02:06 Motivation for Spatial Interpolation 03:13 Spatial Interpolation 04:13 1-D Example: Motivation 06:31 1-D Example: Inference on New Data 08:42 1-D Example: Inference on New Data 09:51 Gaussian Process (GP) 10:40 Covariance (Kernal) for GPR 12:10 GPR Workflow 14:00 Filtered Kriing Lab Demo 22:49 Spatial Interpolation Based on GPR 25:04 Spatial Interpolation Based on GPR 25:26 Spatial Interpolation Based on GPR 26:08 Spatial Interpolation Based on GPR 26:26 Conventional GPR-Based Methods 27:21 Filtered Kriging 28:17 Improved Covariance Modeling with FK 29:38 Improved Covariance Modeling with FK 30:39 Improved Covariance Modeling with FK 31:54 Improved Covariance Modeling with FK 33:29 Tutorial to Filtered Kriging for Spatial Interpolaton This presentation and related downloads can be found at: https://nanohub.org/resources/36189