A Fast Non-Negative Latent Factor Model Based on Generalized Momentum Method

Non-negative latent factor (NLF) models can efficiently acquire useful knowledge from high-dimensional and sparse (HiDS) matrices filled with non-negative data. Single factor-dependent, multiplicative update (SLF-NMU) is an efficient algorithm for building NLF model on HiDS matrix, yet it suffers slow convergence. A momentum method frequently adopted to accelerate a learning algorithm, but incompatible those implicitly adopting gradients like SLF-NMU. To build fast (FNLF) model, we propose generalized compatible With it, further single factor-dependent non-negative, momentum-incorporated thereby achieving FNLF model. Empirical studies six industrial application indicate that outperforms in terms of both convergence rate prediction accuracy missing Hence, compared more practical applications.