CBMAP: Clustering-Based Manifold Approximation and Projection for Dimensionality Reduction

Basit Öğe Kaydı
dc.contributor.authorDogan, Berat
dc.date.accessioned2026-04-04T13:33:24Z
dc.date.available2026-04-04T13:33:24Z
dc.date.issued2025
dc.departmentİnönü Üniversitesi
dc.description.abstractDimensionality reduction methods are employed to decrease data dimensionality, either to enhance machine learning performance or to facilitate data visualization in two or three-dimensional spaces. Among these, popular nonlinear methods such as t-SNE, UMAP, TriMap, and PaCMAP excel in capturing local relationships and nonlinear structures. However, they often distort the global arrangement of clusters, rely heavily on hyperparameter tuning, and exhibit sensitivity to initialization. Moreover, most of these methods cannot project unseen test samples, limiting their applicability in real-world scenarios. To address these challenges, this study introduces a novel approach, CBMAP (Clustering-Based Manifold Approximation and Projection), which explicitly incorporates clustering in the high-dimensional space to guide the embedding. CBMAP computes membership values based on cluster centers in the original space and preserves these memberships during the projection process. This design enables CBMAP to better retain the global layout of the data while maintaining meaningful local relationships. CBMAP demonstrates low sensitivity to initialization strategies, minimal dependence on hyperparameters, and supports projection of unseen test samples. Experimental evaluations on both toy and real-world benchmark datasets show that CBMAP consistently preserves global structures and inter-cluster distances more effectively than state-of-the-art methods, while delivering competitive results in local structure preservation. The method is freely available at https://github.com/doganlab/cbmap and can be installed via the Python Package Index with the command pip install cbmap.
dc.description.sponsorshipInonu University Scientific Research Projects Coordination Unit [FBA-2025-4100]; Scientific and Technological Research Council of Turkiye (TUBIdot;TAK) [120C152]
dc.description.sponsorshipThis work was supported in part by the Inonu University Scientific Research Projects Coordination Unit under Project FBA-2025-4100, and in part by the Scientific and Technological Research Council of Turkiye (TUB & Idot;TAK) under Project 120C152.
dc.identifier.doi10.1109/ACCESS.2025.3599722
dc.identifier.endpage145167
dc.identifier.issn2169-3536
dc.identifier.orcid0000-0003-4810-1970
dc.identifier.scopus2-s2.0-105013761426
dc.identifier.scopusqualityQ1
dc.identifier.startpage145158
dc.identifier.urihttps://doi.org/10.1109/ACCESS.2025.3599722
dc.identifier.urihttps://hdl.handle.net/11616/109136
dc.identifier.volume13
dc.identifier.wosWOS:001556087800019
dc.identifier.wosqualityQ2
dc.indekslendigikaynakWeb of Science
dc.indekslendigikaynakScopus
dc.institutionauthorDogan, Berat
dc.language.isoen
dc.publisherIeee-Inst Electrical Electronics Engineers Inc
dc.relation.ispartofIEEE Access
dc.relation.publicationcategoryMakale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı
dc.rightsinfo:eu-repo/semantics/openAccess
dc.snmzKA_WOS_20250329
dc.subjectClustering algorithms
dc.subjectDimensionality reduction
dc.subjectData visualization
dc.subjectPrincipal component analysis
dc.subjectMachine learning algorithms
dc.subjectHigh dimensional data
dc.subjectSensitivity
dc.subjectManifolds
dc.subjectData structures
dc.subjectStandards
dc.subjectClustering
dc.subjectdimensionality reduction
dc.subjectk-means
dc.subjectPCA
dc.subjectt-SNE
dc.subjectUMAP
dc.titleCBMAP: Clustering-Based Manifold Approximation and Projection for Dimensionality Reduction
dc.typeArticle

Dosyalar

Koleksiyon

WoS İndeksli Yayınlar Koleksiyonu
Scopus İndeksli Yayınlar Koleksiyonu