Speed-up and multi-view extensions to subclass discriminant analysis

In this paper, we propose a speed-up approach for subclass discriminant analysis and formulate novel efficient multi-view solution to it. The is developed based on graph embedding spectral regression approaches that involve eigendecomposition of the corresponding Laplacian matrix its eigenvectors. We show by exploiting structure between-class matrix, step can be substituted with much faster process. Furthermore, criterion an it obtained in similar single-view manner. evaluate proposed methods nine datasets compare them related existing approaches. Experimental results solutions achieve competitive performance, often outperforming methods. At same time, they significantly decrease training time.