【摘要】 Quasi-Regression主要用于解决高维空间的函数逼近问题.虽然这种方法的计算效率非常高,但拟合精度依赖于标量系数的估计.在正交基的线性组合中,选择Quasi-Regression标量系数的最小方差的无偏估计量,从而得到标量系数的新统计量,使得拟合函数的期望由原来的O(pn)减少到O(pn2).更多还原

【Abstract】 Quasi-Regression is introduced for approximation of a function in high dimensional space in recent literature.It was very highly computationally efficient.But the accuracy of approximation to an unknown function depends on estimators of scalar coefficients.In this paper,the authors provide new estimators of scalar coefficients,and obtain more accurate fitted function of an unknown function than the usual ones.更多还原