A Scalable Unsupervised Feature Selection With Orthogonal Graph Representation for Hyperspectral Images

dc.authorid Taskin, Gulsen/0000-0002-2294-4462
dc.authorid Camps-Valls, Gustau/0000-0003-1683-2138
dc.authorwosid Taskin, Gulsen/ABI-7693-2020
dc.authorwosid Camps-Valls, Gustau/A-2532-2011
dc.contributor.author Yetkin, Emrullah Fatih
dc.contributor.author Yetkin, E. Fatih
dc.contributor.author Camps-Valls, Gustau
dc.date.accessioned 2023-10-19T15:11:56Z
dc.date.available 2023-10-19T15:11:56Z
dc.date.issued 2023
dc.department-temp [Taskin, Gulsen] Istanbul Tech Univ, Inst Disaster Management, TR-34469 Istanbul, Turkiye; [Yetkin, E. Fatih] Kadir Has Univ, Management Informat Syst Dept, TR-34083 Istanbul, Turkiye; [Camps-Valls, Gustau] Univ Valencia, Image Proc Lab, Valencia 46010, Spain en_US
dc.description.abstract Feature selection (FS) is essential in various fields of science and engineering, from remote sensing to computer vision. Reducing data dimensionality by removing redundant features and selecting the most informative ones improves machine learning algorithms' performance, especially in supervised classification tasks, while lowering storage needs. Graph-embedding (GE) techniques have recently been found efficient for FS since they preserve the geometric structure of the original feature space while embedding data into a low-dimensional subspace. However, the main drawback is the high computational cost of solving an eigenvalue decomposition problem, especially for large-scale problems. This article addresses this issue by combining the GE framework and representation theory for a novel FS method. Inspired by the high-dimensional model representation (HDMR), the feature transformation is assumed to be a linear combination of a set of univariate orthogonal functions carried out in the GE framework. As a result, an explicit embedding function is created, which can be utilized to embed out-of-samples into low-dimensional space and provide a feature relevance score. The significant contribution of the proposed method is to divide an $n$ -dimensional generalized eigenvalue problem into $n$ small-sized eigenvalue problems. With this property, the computational complexity (CC) of the GE is significantly reduced, resulting in a scalable FS method, which could be easily parallelized too. The performance of the proposed method is compared favorably to its counterparts in high-dimensional hyperspectral image (HSI) processing in terms of classification accuracy, feature stability, and computational time. en_US
dc.description.sponsorship Scientific and Technological Research Council of Turkey (TUEBITAK) [217E032, 1001]; European Research Council (ERC) through the ERC Synergy Grant Project Understanding and Modeling the Earth System with Machine Learning (USMILE) [855187] en_US
dc.description.sponsorship This work was supported by the Scientific and Technological Research Council of Turkey (TUEBITAK)-1001 under Project 217E032. The work of Gustau Camps-Valls was supported by the European Research Council (ERC) through the ERC Synergy Grant Project Understanding and Modeling the Earth System with Machine Learning (USMILE) under Grant 855187.& nbsp; en_US
dc.identifier.citationcount 1
dc.identifier.doi 10.1109/TGRS.2023.3284475 en_US
dc.identifier.issn 0196-2892
dc.identifier.issn 1558-0644
dc.identifier.scopus 2-s2.0-85162732832 en_US
dc.identifier.scopusquality Q1
dc.identifier.uri https://doi.org/10.1109/TGRS.2023.3284475
dc.identifier.uri https://hdl.handle.net/20.500.12469/5280
dc.identifier.volume 61 en_US
dc.identifier.wos WOS:001021331900015 en_US
dc.identifier.wosquality Q1
dc.khas 20231019-WoS en_US
dc.language.iso en en_US
dc.publisher IEEE-Inst Electrical Electronics Engineers Inc en_US
dc.relation.ispartof Ieee Transactions on Geoscience and Remote Sensing en_US
dc.relation.publicationcategory Makale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı en_US
dc.rights info:eu-repo/semantics/closedAccess en_US
dc.scopus.citedbyCount 4
dc.subject Band Selection En_Us
dc.subject Index Terms- Dimensionality reduction en_US
dc.subject Classification En_Us
dc.subject feature selection (FS) en_US
dc.subject global sensitivity analysis en_US
dc.subject Band Selection
dc.subject graph embedding (GE) en_US
dc.subject Classification
dc.subject hyperspectral image (HSI) analysis en_US
dc.title A Scalable Unsupervised Feature Selection With Orthogonal Graph Representation for Hyperspectral Images en_US
dc.type Article en_US
dc.wos.citedbyCount 2
dspace.entity.type Publication
relation.isAuthorOfPublication 81114204-31da-4513-a19f-b5446f8a3a08
relation.isAuthorOfPublication.latestForDiscovery 81114204-31da-4513-a19f-b5446f8a3a08

Files

Original bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
5280.pdf
Size:
8.89 MB
Format:
Adobe Portable Document Format
Description:
Tam Metin / Full Text