Life of Inches

版权声明：转载时请以超链接形式标明文章原始出处和作者信息及本声明
http://anthonychan.blogbus.com/logs/31512974.html

For a matrix M containing the information we need, we can always rewrite M as

M = XM`X^T

Where X is the eigenvector of M and M` is a diagonal matrix whose elements are the eigenvalues of M. To decrease the dimension of M, we can select part of M` to form a smaller matrix M`` where only some significant eigenvalues remain.

*A similar application of SVD is its version in text analysis where it is renamed as Latent Semantic Indexing.