人脸识别的核方法研究

Research on nuclear method for face recognition

基于核方法的子空间分析方法在高维空间中有解决非线性问题的优势,在提出的LMDA和流形上的线性回归方法的基础上,把LMDA算法拓展到高维空间,提出了基于核的LMDA算法,同时,由表达定理出发,用核回归方法估计数据与流形上的局部坐标之间的变换矩阵,推出了流形上的非线性回归算法。

the kernel method for subspace analysis method has the advantage to solve nonlinear problems in high dimensional space based on the basis of linear, in the proposed LMOA and manifold regressionmethod, the LMDA algorithm is extended to high dimension space, LMDA algorithm is proposed, based on core at the same time, starting from the expression theorem, transformation matrix between local coordinate estimation data and kernel regression method on manifolds, launched on nonlinear manifold regression algorithm.