Autoencoder-based Low-Rank Spectral Ensemble Clustering of Biological Data

This work presents a cluster ensemble algorithm using a combination of low-rank co-association matrix decomposition, deep autoencoder transformation, and spectral clustering. The suggested algorithm is studied on Mice Protein Expression dataset and Cardiotocography dataset. Monte-Carlo simulations are used to evaluate the clustering performance. The experiments show that the proposed algorithm significantly outperforms other considered variants of clustering framework with respect to clustering accuracy.