摘要:Non-negative Matrix Factorization (NMF) is a well established dimension reduction technique. NMF reduces the dimension of the non-negative data matrix by applying the non- negativity constraint. One of the applications of NMF is recognition problem (e-g face recognition). To solve recognition problem using NMF, the order of decomposition play an important role on the recognition rate. This paper presents the details analysis of the order of decomposition on the recognition rate. A standard face dataset A and T is used for analysis.