The Minimum Volume Ellipsoid Metric

The Minimum Volume Ellipsoid Metric
复制标题

最小体积椭球度量

DOI:
--
复制
发表时间:
2007
期刊:
DAGM-Symposium
影响因子:
--
通讯作者:
F. Ferrie
F. Ferrie
中科院分区:
--
文献类型:
--
作者:
Karim Abou;F. Ferrie

文献摘要

被引文献

相似文献

我们提出了一个无监督的“本地学习”算法学习的输入空间中的度量。几何上,对于一个给定的查询点,该算法找到覆盖其邻域的最小体积椭球(MVE),表征其邻域变量的相关性和方差。在代数上,该算法最大化局部协方差矩阵的行列式,这相当于一个凸优化问题。最后的矩阵参数化的马氏度量产生的MVE度量(MVEM)。该指标进行了测试,在监督学习任务,并显示出有前途的和有竞争力的结果相比,在文献中的最先进的指标。
We propose an unsupervised "local learning" algorithm for learning a metric in the input space. Geometrically, for a given query point, the algorithm finds the minimum volume ellipsoid (MVE) covering its neighborhood which characterizes the correlations and variances of its neighborhood variables. Algebraically, the algorithm maximizes the determinant of the local covariance matrix which amounts to a convex optimization problem. The final matrix parameterizes a Mahalanobis metric yielding the MVE metric (MVEM). The proposed metric was tested in a supervised learning task and showed promising and competitive results when compared with state of the art metrics in the literature.