一种平滑的张量分解算法研究

Research on a Smooth Tensor Decomposition Algorithm

张量方法为高维数据提供了有效的分析方法。提出了一种基于图正则化和Lp平滑约束的非负Tucker分解方法,结合各向同性(L2范数)和各向异性(L1范数)扩散平滑的优点,并产生优化问题的平滑和更精确的解,通过实验验证了该模型的有效性。

Tensor method provides an effective analysis method for high-dimensional data.This paper proposes a nonnegative Tucker decomposition method based on graph regularization and Lp smoothing constraint,which combines the advantages of isotropic(L2 norm)and anisotropic(L1 norm)diffusion smoothing,and obtains a smooth and more accurate solution of the optimization problem.The effectiveness of the model is verified by experiments.