A New Distance Metric Learning Algorithm
A New Distance Metric Learning Algorithm
复制标题
DOI：
--
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
Zhang Li-ming
中科院分区：
文献类型：
--
作者：
Zhang Li-ming
提出了一种新的距离度量学习算法，通过学习最大化类间对的最小距离和最小化类内对的最大距离，用一个可微凸函数逼近目标函数，避免了局部极小值，同时，该方法也可用于线性降维，通过与其他距离度量的比较，实验结果验证了该方法的有效性。几个人脸数据库
A new distance metric learning algorithm is proposed.By learning maximizing the minimal distance of inter-class pairs and minimizing the maximal distance of intra-class pairs simultaneously,a differentiable convex function is used to approximate our objective function,thus avoid local minima.Besides,the method can also be used in linear dimensionality reduction.Experimental results validate the effectiveness of our method by comparing with other distance metrics on several face databases.