An efficient nonnegative matrix factorization approach in flexible kernel space
Daoqiang Zhang, Wanquan Liu · eSpace (Curtin University) · 2009
In this paper, we propose a general formulation for kernel nonnegative matrix factorization with flex-ible kernels. Specifically, we propose the Gaus-sian nonnegative matrix factorization (GNMF) al-gorithm by using the Gaussian kernel in the frame-work. Different from a recently developed polyno-mial NMF (PNMF), GNMF finds basis vectors in the kernel-induced feature space and the computa-tional cost is independent of input dimensions. Fur-thermore, we prove the convergence and nonnega-tivity of decomposition of our method. Extensive experiments compared with PNMF and other NMF algorithms on several face databases, validate the effectiveness of the proposed method. 1