Scale and curvature effects in principal geodesic analysis

Scale and curvature effects in principal geodesic analysis
复制标题

主测地线分析中的尺度和曲率效应

DOI:
10.1016/j.jmva.2016.09.009
复制
发表时间:
2017
影响因子:
1.6
通讯作者:
Lin, Lizhen
Lin, Lizhen
中科院分区:
数学2区
文献类型:
--
作者:
Lazar, Drew;Lin, Lizhen

文献摘要

参考文献

被引文献

相似文献

利用微分几何和统计学之间的密切联系来建模光滑流形值数据的兴趣越来越大。特别是,最近已经做了大量工作将主成分分析(PCA)(线性空间中的降维方法)推广到Riemannian流形。一种这样的推广被称为主测地线分析(PGA)。本文以一种新的方式,获得了PGA目标函数域中引入的尺度参数的泰勒展开式。结果表明,这种技术不仅导致更好的封闭形式的PGA的近似,但也揭示了规模,曲率和分布的数据上的PGA的解决方案和他们的差异,一阶切空间近似的效果。这种方法应该能够不仅适用于PGA,但也适用于其他推广的PCA和更普遍的其他内在统计黎曼流形。
There is growing interest in using the close connection between differential geometry and statistics to model smooth manifold-valued data. In particular, much work has been done recently to generalize principal component analysis (PCA), the method of dimension reduction in linear spaces, to Riemannian manifolds. One such generalization is known as principal geodesic analysis (PGA). This paper, in a novel fashion, obtains Taylor expansions in scaling parameters introduced in the domain of objective functions in PGA. It is shown this technique not only leads to better closed-form approximations of PGA but also reveals the effects that scale, curvature and the distribution of data have on solutions to PGA and on their differences to first-order tangent space approximations. This approach should be able to be applied not only to PGA but also to other generalizations of PCA and more generally to other intrinsic statistics on Riemannian manifolds.
使用方向统计来研究人体运动学的变化
DOI: --
发表时间: 2000
期刊:
影响因子: --
作者:
Denis Rancourt;Louis;Jérôme Asselin
通讯作者: Jérôme Asselin
黎曼流形的本征多元方差分析及其在肯德尔平面形状空间中的应用
DOI: 10.1109/tpami.2009.117
发表时间: 2010
影响因子: 23.6
作者:
S. Huckemann;T. Hotz;A. Munk
通讯作者: A. Munk
旋转:SO(3) 数据的 R 包
DOI: --
发表时间: 2014
期刊: The R Journal
影响因子: --
作者:
B. Stanfill;H. Hofmann;Ulrike Genschel
通讯作者: Ulrike Genschel
DOI: --
发表时间: 2012
期刊:
影响因子: --
作者:
J. Eschenburg
通讯作者: J. Eschenburg
SO(3) 数据的精确主测地线分析
DOI: --
发表时间: 2007
期刊: European Signal Processing Conference
影响因子: --
作者:
S. Said;N. Courty;N. L. Bihan;S. Sangwine
通讯作者: S. Sangwine