Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of Glioma

In the era of big data, high-dimensional datasets have become increasingly common in fields such as biometrics, computer vision, and medical imaging. While such data contain abundant information, they are often accompanied by substantial noise, high redundancy, and complex intrinsic structures, posi...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhichao Song, Meiling Chen, Liang Xie, Xi Fang
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/15/11/6177
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849722571509989376
author Zhichao Song
Meiling Chen
Liang Xie
Xi Fang
author_facet Zhichao Song
Meiling Chen
Liang Xie
Xi Fang
author_sort Zhichao Song
collection DOAJ
description In the era of big data, high-dimensional datasets have become increasingly common in fields such as biometrics, computer vision, and medical imaging. While such data contain abundant information, they are often accompanied by substantial noise, high redundancy, and complex intrinsic structures, posing significant challenges for analysis and modeling. To address these issues, unsupervised feature selection has attracted growing interest due to its ability to handle unlabeled, noisy, and unstructured data. This paper proposes a novel unsupervised feature selection algorithm based on a dual-graph autoencoder (DGA), which combines the powerful data reconstruction capability of autoencoders with the structural preservation strengths of graph regularization. Specifically, the algorithm introduces the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><msub><mi mathvariant="script">l</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn><mo>/</mo><mn>2</mn></mrow></msub></mrow></semantics></math></inline-formula>-norm and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><msub><mi mathvariant="script">l</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn></mrow></msub></mrow></semantics></math></inline-formula>-norm constraints on the encoder and decoder weight matrices, respectively, to promote feature sparsity and suppress redundancy. Furthermore, an <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><msub><mi mathvariant="script">l</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn><mo>/</mo><mn>2</mn></mrow></msub></mrow></semantics></math></inline-formula>-norm loss term is introduced to enhance robustness against noise and outliers. Two separate adjacency graphs are constructed to capture the local geometric relationships among samples and among features, and their corresponding graph regularization terms are embedded in the training process to retain the intrinsic structure of the data. Experiments on multiple benchmark datasets and [<sup>68</sup>Ga]Ga-Pentixafor PET/CT glioma imaging data demonstrate that the proposed DGA significantly improves clustering performance and accurately identifies features associated with lesion regions. From a clinical perspective, DGA facilitates more accurate lesion characterization and biomarker identification in glioma patients, thereby offering potential utility in aiding diagnosis, treatment planning, and personalized prognosis assessment.
format Article
id doaj-art-ed87b0d06fcd4ee2ac12fa482729fe02
institution DOAJ
issn 2076-3417
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Applied Sciences
spelling doaj-art-ed87b0d06fcd4ee2ac12fa482729fe022025-08-20T03:11:18ZengMDPI AGApplied Sciences2076-34172025-05-011511617710.3390/app15116177Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of GliomaZhichao Song0Meiling Chen1Liang Xie2Xi Fang3School of Mathematics and Statistics, Wuhan University of Technology, Wuhan 430070, ChinaSchool of Mathematics and Statistics, Wuhan University of Technology, Wuhan 430070, ChinaSchool of Mathematics and Statistics, Wuhan University of Technology, Wuhan 430070, ChinaSchool of Mathematics and Statistics, Wuhan University of Technology, Wuhan 430070, ChinaIn the era of big data, high-dimensional datasets have become increasingly common in fields such as biometrics, computer vision, and medical imaging. While such data contain abundant information, they are often accompanied by substantial noise, high redundancy, and complex intrinsic structures, posing significant challenges for analysis and modeling. To address these issues, unsupervised feature selection has attracted growing interest due to its ability to handle unlabeled, noisy, and unstructured data. This paper proposes a novel unsupervised feature selection algorithm based on a dual-graph autoencoder (DGA), which combines the powerful data reconstruction capability of autoencoders with the structural preservation strengths of graph regularization. Specifically, the algorithm introduces the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><msub><mi mathvariant="script">l</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn><mo>/</mo><mn>2</mn></mrow></msub></mrow></semantics></math></inline-formula>-norm and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><msub><mi mathvariant="script">l</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn></mrow></msub></mrow></semantics></math></inline-formula>-norm constraints on the encoder and decoder weight matrices, respectively, to promote feature sparsity and suppress redundancy. Furthermore, an <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><msub><mi mathvariant="script">l</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn><mo>/</mo><mn>2</mn></mrow></msub></mrow></semantics></math></inline-formula>-norm loss term is introduced to enhance robustness against noise and outliers. Two separate adjacency graphs are constructed to capture the local geometric relationships among samples and among features, and their corresponding graph regularization terms are embedded in the training process to retain the intrinsic structure of the data. Experiments on multiple benchmark datasets and [<sup>68</sup>Ga]Ga-Pentixafor PET/CT glioma imaging data demonstrate that the proposed DGA significantly improves clustering performance and accurately identifies features associated with lesion regions. From a clinical perspective, DGA facilitates more accurate lesion characterization and biomarker identification in glioma patients, thereby offering potential utility in aiding diagnosis, treatment planning, and personalized prognosis assessment.https://www.mdpi.com/2076-3417/15/11/6177unsupervised feature selectionautoencoderdual-graph regularizationnorm[<sup>68</sup>Ga]Ga-Pentixafor PET
spellingShingle Zhichao Song
Meiling Chen
Liang Xie
Xi Fang
Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of Glioma
Applied Sciences
unsupervised feature selection
autoencoder
dual-graph regularization
norm
[<sup>68</sup>Ga]Ga-Pentixafor PET
title Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of Glioma
title_full Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of Glioma
title_fullStr Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of Glioma
title_full_unstemmed Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of Glioma
title_short Unsupervised Feature Selection via a Dual-Graph Autoencoder with <inline-formula><math display="inline"><semantics><mrow><msub><mi mathvariant="bold-script">l</mi><mrow><mn mathvariant="bold">2</mn><mo mathvariant="bold">,</mo><mn mathvariant="bold">1</mn><mo mathvariant="bold">/</mo><mn mathvariant="bold">2</mn></mrow></msub></mrow></semantics></math></inline-formula>-Norm for [<sup>68</sup>Ga]Ga-Pentixafor PET Imaging of Glioma
title_sort unsupervised feature selection via a dual graph autoencoder with inline formula math display inline semantics mrow msub mi mathvariant bold script l mi mrow mn mathvariant bold 2 mn mo mathvariant bold mo mn mathvariant bold 1 mn mo mathvariant bold mo mn mathvariant bold 2 mn mrow msub mrow semantics math inline formula norm for sup 68 sup ga ga pentixafor pet imaging of glioma
topic unsupervised feature selection
autoencoder
dual-graph regularization
norm
[<sup>68</sup>Ga]Ga-Pentixafor PET
url https://www.mdpi.com/2076-3417/15/11/6177
work_keys_str_mv AT zhichaosong unsupervisedfeatureselectionviaadualgraphautoencoderwithinlineformulamathdisplayinlinesemanticsmrowmsubmimathvariantboldscriptlmimrowmnmathvariantbold2mnmomathvariantboldmomnmathvariantbold1mnmomathvariantboldmomnmathvariantbold2mnmrowmsubmrowsemanticsmat
AT meilingchen unsupervisedfeatureselectionviaadualgraphautoencoderwithinlineformulamathdisplayinlinesemanticsmrowmsubmimathvariantboldscriptlmimrowmnmathvariantbold2mnmomathvariantboldmomnmathvariantbold1mnmomathvariantboldmomnmathvariantbold2mnmrowmsubmrowsemanticsmat
AT liangxie unsupervisedfeatureselectionviaadualgraphautoencoderwithinlineformulamathdisplayinlinesemanticsmrowmsubmimathvariantboldscriptlmimrowmnmathvariantbold2mnmomathvariantboldmomnmathvariantbold1mnmomathvariantboldmomnmathvariantbold2mnmrowmsubmrowsemanticsmat
AT xifang unsupervisedfeatureselectionviaadualgraphautoencoderwithinlineformulamathdisplayinlinesemanticsmrowmsubmimathvariantboldscriptlmimrowmnmathvariantbold2mnmomathvariantboldmomnmathvariantbold1mnmomathvariantboldmomnmathvariantbold2mnmrowmsubmrowsemanticsmat