摘要
高维数据降维方法已经被广泛应用在信息检索、模式识别、数据挖掘和人工智能等领域。针对目前流形学习方法的嵌入效果非常敏感于局部邻域的选取方式,提出一种自适应邻域选择的局部线性降维方法。该方法评估真实数据的固有维数,判断每一数据点的局部切方向,以便自适应地选择每一数据点的邻域数,使得不同数据集与邻域选取方式之间存在很好的自适应性,实现更好的降维效果。在人工生成数据集和医学数据上的仿真结果表明,该方法起到了良好的降维效果。
Dimensional reduction method for high-dimensional data have been widely used in many areas such as information retrieval,pattern recognition,data mining and artificial intelligence,etc.The embedded results of existing manifold learning methods are very sensitive to the selection mode of local neighbours,to address this issue,we propose a local linear dimensional reduction method which is based on adaptive neighbourhood selection.The method estimates the intrinsic dimensionality of real data and determines local tangent orientation of each data point in order to adaptively select the neighbourhood of each data point,this enables a satisfied self-adaptability between the different data sets and the neighbourhood selection mode,and achieves better dimensional reduction effect.Simulation results show that the proposed method achieves a pretty good dimensional reduction effect on artificially generated data sets and medical data.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第4期255-257,共3页
Computer Applications and Software
基金
宜宾学院校级青年基金项目(2010Q37)
关键词
高维数据降维
流形学习
局部邻域
固有维数
局部切方向
Dimensional reduction of high-dimensional data Manifold learning Local neighbourhood Intrinsic dimensionality Local tangent orientation