Research On Incremental Non-negative Matrix Factorization Algorithm

Non-negative Matrix Factorization algorithm is a recently developed and studied technique for finding linear representations of non-negative data.When existing Non-negative Matrix Factorization algorithm applies to a problem of incremental scale,the consumption of space and time behaves inefficiency.In this paper,an Incremental Nonnegative Matrix Factorization algorithm is proposed,which uses partitioned matrix theory to reduce the computing scale,and uses decomposition results already derived to avoid re-calculating every time.We analyze the properties of this incremental algorithm,prove its correctness and compare it to the similar algorithms.Experiments show that the new algorithm performs efficient for saving computing resources.