Simulační algoritmus SVD pro efektivní analýzu rozsáhlých dat

Abstract

This bachelor thesis deal with effective application Monte-Carlo Method over SVD decomposition. SVD decomposition is an effective and robust decomposotion method of linear algebra. The application of this decomposition is in all computers sciences. For example, compression of pictures, the SVD is used in the PCA method, to analysis for ex. to human faces. Common computing of SVD is demanding to CPU and RAM, primarily in case of very large matrix. But multimedia data contains often unimportant information, so we dont have to compute an approximation of the SVD of complete matrix. We applicate Monte-Carlo method, with this method we sucessfully reduce the size of the matrix.