Cluster Forests

Abstract

With inspiration from Random Forests (RF) in the context of classification, a new clustering ensemble method---Cluster Forests (CF) is proposed. Geometrically, CF randomly probes a high-dimensional data cloud to obtain "good local clusterings" and then aggregates via spectral clustering to obtain cluster assignments for the whole dataset. The search for good local clusterings is guided by a cluster quality measure kappa. CF progressively improves each local clustering in a fashion that resembles the tree growth in RF. Empirical studies on several real-world datasets under two different performance metrics show that CF compares favorably to its competitors. Theoretical analysis reveals that the kappa measure makes it possible to grow the local clustering in a desirable way---it is "noise-resistant". A closed-form expression is obtained for the mis-clustering rate of spectral clustering under a perturbation model, which yields new insights into some aspects of spectral clustering.

23 pages, 6 figures

References in corpus (5)

• Mixed membership stochastic blockmodels
• Consistency of spectral clustering
• Experiments with Random Projection
• Diffusion Maps, Spectral Clustering and Eigenfunctions of Fokker-Planck operators
• Multiway Spectral Clustering: A Margin-Based Perspective

Cited by in corpus (6)

• The geometry of kernelized spectral clustering
• Global-scale massive feature extraction from monthly hydroclimatic time series: Statistical characterizations, spatial patterns and hydrological similarity
• Massive feature extraction for explaining and foretelling hydroclimatic time series forecastability at the global scale
• A First Course in Data Science
• K-nearest Neighbor Search by Random Projection Forests
• Fast communication-efficient spectral clustering over distributed data