low rank approximation
definition
- finding matrix of lower rank, close to a given matrix
- eg. given m x n data matrix, seek a rank-k approximation, where k << m,n
- useful for reducing the complexity & storage of data, or extracting most important features from data (refer to PCA)