Rights statement: This is the author’s version of a work that was accepted for publication in Pattern Recognition. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Pattern Recognition, 107, 2020 DOI: 10.1016/j.patcog.2020.107508
Accepted author manuscript, 1.53 MB, PDF document
Available under license: CC BY-NC-ND: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License
Final published version
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Nonlinear Dimensionality Reduction for Clustering
AU - Tasoulis, Sotiris
AU - Pavlidis, Nicos
AU - Roos, Teemu
N1 - This is the author’s version of a work that was accepted for publication in Pattern Recognition. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Pattern Recognition, 107, 2020 DOI: 10.1016/j.patcog.2020.107508
PY - 2020/11/1
Y1 - 2020/11/1
N2 - We introduce an approach to divisive hierarchical clustering that is capable of identifying clusters in nonlinear manifolds. This approach uses the isometric mapping (Isomap) to recursively embed (subsets of) the data in one dimension, and then performs a binary partition designed to avoid the splitting of clusters. We provide a theoretical analysis of the conditions under which contiguous and high density clusters in the original space are guaranteed to be separable in the one dimensional embedding. To the best of our knowledge there is little prior work that studies this problem. Extensive experiments on simulated and real data sets show that hierarchical divisive clustering algorithms derived from this approach are effective.
AB - We introduce an approach to divisive hierarchical clustering that is capable of identifying clusters in nonlinear manifolds. This approach uses the isometric mapping (Isomap) to recursively embed (subsets of) the data in one dimension, and then performs a binary partition designed to avoid the splitting of clusters. We provide a theoretical analysis of the conditions under which contiguous and high density clusters in the original space are guaranteed to be separable in the one dimensional embedding. To the best of our knowledge there is little prior work that studies this problem. Extensive experiments on simulated and real data sets show that hierarchical divisive clustering algorithms derived from this approach are effective.
KW - Nonlinearity
KW - Dimensionality reduction
KW - Divisive Hierarchical Clustering
KW - Manifold Clustering
U2 - 10.1016/j.patcog.2020.107508
DO - 10.1016/j.patcog.2020.107508
M3 - Journal article
VL - 107
JO - Pattern Recognition
JF - Pattern Recognition
SN - 0031-3203
M1 - 107508
ER -