arXiv Machine Learning
DRtool: An Interactive Tool for Analyzing High-Dimensional Clusterings
Read the original on arXiv Machine Learning →The Flow has not summarised this story yet — read it at arXiv Machine Learning.
Related stories
arXiv Machine Learning
Aug 21
Why Can't I See My Clusters? A Precision-Recall Approach to Dimensionality Reduction Validation
arXiv:2509. 04222v2 Announce Type: replace Abstract: Dimensionality Reduction (DR) is widely used for visualizing high-dimensional data, often with the goal of revealing expected cluster structure.
arXiv Machine Learning
Sep 24
Assessing the impact of dimensionality reduction on clustering performance - a systematic study
The paper systematically evaluates how five dimensionality reduction methods—PCA, Kernel PCA, VAE, Isomap, and MDS—affect the performance of four clustering algorithms (k‑means, AHC, GMM, and OPTICS). Using the Adjusted Rand Index, the study compares clustering quality with and without dimensionality reduction at levels of k‑1, 25%, and 50% of the original dimensions. Results highlight that the choice of reduction technique and its level must be carefully matched to the data’s geometry and the clustering algorithm used.
arXiv Machine Learning
Jul 10
Persistent Multiscale Density-based Clustering
arXiv:2512. 16558v3 Announce Type: replace Abstract: Clustering is a cornerstone of modern data analysis.
arXiv Machine Learning
Jun 2
Cluster Analysis with Resampling for Validation and Exploration (CARVE)
arXiv:2606. 00327v1 Announce Type: cross Abstract: Clustering is widely used across the sciences as the foundation for downstream data-driven scientific discoveries.
arXiv AI
Aug 10
Density-aware Hierarchical Clustering Based on Element-Categorized Connection Subgraphs
arXiv:2608. 06990v1 Announce Type: cross Abstract: Clustering is a fundamental data mining technique for pattern recognition through unsupervised learning.
arXiv AI
Sep 4
Mixed Data Clustering Survey and Challenges
The paper "Mixed Data Clustering Survey and Challenges" discusses how the rise of big data has made clustering of heterogeneous datasets—containing both numerical and categorical variables—particularly difficult for traditional methods. It highlights the importance of hierarchical and explainable algorithms for producing interpretable results that aid decision‑making. The authors propose a new clustering approach based on pretopological spaces and benchmark it against classical numerical clustering algorithms and existing pretopological methods to evaluate its performance in the big data context.