One-step Spectral Clustering based on Self-paced Learning

Abstract Aiming at traditional spectral clustering method still suffers from the following issues: 1) unable to handle the incomplete data, 2) two-step clustering strategies tend to perform poorly due to the heterogeneity between the similarity matrix learning model and the clustering model, 3) constructing the affinity matrix from original data which often contains noises and outliers. To address these issues, this paper proposes a robust one-step clustering method based on self-paced learning. Specifically, this paper first designs a missing value mapping matrix to handle missing data, and then employs self-paced regularizer to sort the importance of samples, reduce the interference of noise and outliers of the model, second, fusing affinity matrix learning and spectral decomposition into one model to achieve the purpose of one-step clustering. Experimental results on 8 real data sets, verified the effectiveness of our proposed one-step spectral clustering method, compared to state-of-the-art methods.