Abstract:For most clustering methods, not only the number of clusters must be set in advance, but also various hyperparameters such as initial centroids, number of nearest neighbours, the minimum number of points, neighbourhood radius, and cutoff distance all require pre-specification. As one of the most promising unsupervised learning methods in machine intelligence, existing clustering methods cannot simultaneously handle datasets with arbitrary shapes, different densities, distinct sizes, and overlapping. Background outliers and high dimensionality make clustering problems more challenging. In this paper, we propose a novel universal clustering methodology, called G2-SCANN, which yields the best clustering performance for all 30 synthetic and real datasets without any hyperparameter tuning if the exact number of clusters is known. Firstly, the shortest path length (SPL) in complex network or graph-based geodesic distance is used to give a locally backbone-structured description of graph vertex similarity. Accordingly, SPL-weighted local degree (SLD) is defined as vertex attributes of a SPL-weighted graph expressed by G2-SPL adjacency matrix with ε-natural neighbourhood. Secondly, the process of calculating SLD for every data point in a bottom-up way directly leads to division from a complete graph constituted by all data points to a group of SLD trees. This brings the interpretability and the elimination of lone trees. Thirdly, contrastive learning of largest SLD values for finding root vertices of each divisive tree is conducted and top-down category message is then transmitted from the root vertices to all the leaf ones of a SLD tree. It eventually produces tree-like clusters. Totally, the proposed G2-SCANN method leverages both local neighbouring similarity of data points and global information about data distribution and makes it perform better than other methods.

A Self-Adaptive Graph-Based Clustering Method with Noise Identification

A robust clustering method with noise identification based on directed K-nearest neighbor graph

Self-Adaptive Clustering of Dynamic Multi-Graph Learning

A Novel Graph-Based Clustering Method Using Noise Cutting

Adaptive Gravitational Clustering Algorithm Integrated with Noise Detection

A Parameterless Clustering Algorithm Based on Strict Neighborhood Graph

Robust Self-Tuning Sparse Subspace Clustering.

G2-SCANN: Gaussian-kernel Graph-Based SLD Clustering Algorithm with Natural Neighbourhood

A Hierarchical Clustering Algorithm Based on Noise Removal

A Self-Adaptive Spectral Clustering Algorithm

An Adaptive Spectral Clustering Algorithm Based on Natural Neighbor

A Dual Adaptive Assignment Approach for Robust Graph-Based Clustering

Saturated Neighborhood Graph Clustering Optimization Algorithm Based on Edge Information

Self-adaption grey DBSCAN clustering

Non-parameter Clustering Algorithm Based on Chain Propagation and Natural Neighbor

Deep Self-Supervised Attributed Graph Clustering for Social Network Analysis

A New Adaptive Elastic Net Method for Cluster Analysis.

Nonlinear Subspace Clustering Via Adaptive Graph Regularized Autoencoder.

A Novel Clustering Method Based on Adjacent Grids Searching

An Improved Density-Based Spatial Clustering of Application with Noise

Noises Cutting and Natural Neighbors Spectral Clustering Based on Coupling P System