核密度估计的随机加权逼近  

Random weighting approach for kernel density estimation

作　　者：倪龙强[1] 张旭阳 高社生[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072 [2]西安艾菲特光电技术有限公司,陕西西安710061

出　　处：《西北大学学报（自然科学版）》2008年第2期225-228,共4页Journal of Northwest University（Natural Science Edition）

基　　金：陕西省自然科学基金资助项目(N6CS0003)

摘　　要：目的证明核密度估计随机加权逼近的有效性。方法用随机加权法对核密度进行估计。结果在适当的条件下,(nh_n)^(1/2)(■_n(x)-fn(x))和(nh_n)^(1/2)(f_n(x)-f(x))对几乎所有的样本序列X_1,X_2…具有相同的极限分布,同时,得到了随机加权逼近的收敛速度。结论用随机加权法对核密度进行估计是可行的。Aim To prove the validity of random weighting estimation approximation for kernel density. Methods Use the random weighting methods to estimate the kernel density. Results The results show that under certain condition,√nhn（Hn（x）-fn（x））and √nhn（fn（x）-f（x））have a same Limit Distribution for almost all series X1 ,X2…, also the speed of random weighting approximation is gained. Conclusion Form the study the conclusion has got that it is completely viable to estimate the kernel density with the random weighting methods.

关 键 词：随机加权法 核密度估计 BOOTSTRAP逼近 

分 类 号：O212.7[理学—概率论与数理统计]