Robust Graph Regularized Extreme Learning Machine Auto Encoder and Its Application to Single-Cell Samples Classification

Abstract

Combined with Auto Encoder (AE), Extreme Learning Machine Auto Encoder (ELM-AE) has attracted the interest of researchers in recent years. Considering the classification tasks of single-cell Ribonucleic Acid sequencing (scRNA-seq) data, in this paper, we propose a novel supervised learning method based on ELM-AE, which is named Robust Graph Regularized Extreme Learning Machine Auto Encoder (RGELMAE). The method introduces L2,1-norm minimization on loss function to improve the robustness, and combines with the manifold regularization framework to explore the internal local structure between data points. Finally, RGELMAE is applied to the classification tasks of scRNA-seq data. The experimental results indicate that our method can effectively extract the key information representing the original data, and improve the classification performance of ELM.