AutoCluster: Meta-learning Based Ensemble Method for Automated Unsupervised Clustering

Automated clustering automatically builds appropriate clustering models. The existing automated clustering methods are widely based on meta-learning. However, it still faces specific challenges: lacking comprehensive meta-features for meta-learning and general clustering validation index (CVI) as objective function. Therefore, we propose a novel automated clustering method named AutoCluster to address these problems, which is mainly composed of Clustering-oriented Meta-feature Extraction (CME) and Multi-CVIs Clustering Ensemble Construction ((MCEC)-E-2). CME captures the meta-features from spatial randomness and different learning properties of clustering algorithms to enhance meta-learning. (MCEC)-E-2 develops a collaborative mechanism based on clustering ensemble to balance the measuring criterion of different CVIs and construct more appropriate clustering model for given datasets. Extensive experiments are conducted on 150 datasets from OpenML to create meta-data and 33 test datasets from three clustering benchmarks to validate the superiority of AutoCluster. The results show the superiority of AutoCluster for building an appropriate clustering model compared with classical clustering algorithms and CASH method.

Automated summary

AutoCluster is a novel automated clustering method that addresses the challenges of lacking comprehensive meta-features for meta-learning and a general clustering validation index (CVI) as an objective function. It consists of Clustering-oriented Meta-feature Extraction (CME) and Multi-CVIs Clustering Ensemble Construction ((MCEC)-E-2) to enhance meta-learning and balance different CVIs for constructing appropriate clustering models. Extensive experiments show the superiority of AutoCluster compared to classical clustering algorithms and CASH method.