基于子空间插值的领域适应学习  

A domain adaptive learning based on subspace interpolation

作　　者：祁昌平[1] 高彩霞[1] 方媛[1] 

机构地区：[1]河西学院信息技术与传媒学院,甘肃张掖734000

出　　处：《西北师范大学学报（自然科学版）》2014年第5期40-43,共4页Journal of Northwest Normal University(Natural Science)

基　　金：国家自然科学基金资助项目(61262056)

摘　　要：为了寻找一种更好的目标分类器,提出了基于子空间插值的领域适应学习方法.从源域出发,依次训练中间领域字典,缩小样本分布差,形成一条平滑的虚拟路径,逐步适应目标数据,达到提高识别性能的目的.在CMU-PIE上进行了实验,结果表明此方法较其他领域适应学习方法有更好的分类性能.In order to find a better target classifier, this paper proposed the spatial interpolation of domain adaptation method. Starting from the source domain, followed by training intermediate dictionary in the field to further reduce the sampling distribution difference, forming a smooth virtual path, gradually adapting to the target data, achieve the goal of a better performance. And conducted experiment results on the CMU-PIE show that the performance of this method is better than the other areas of adaptive learning method.

关 键 词：子空间插值 领域适应 人脸识别 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]