家家的专栏

原创 latex 段间距离太大

latex 段间距离太大\vspace{-2mm}可在段尾，或者段头加此命令进行调整。

原创 matlab 如何统计计算时间 tic和toc指令

A=rand(100); tic; invA=inv(A); toc;Elapsed time is 0.405422 seconds.

原创 matlab图转pdf，缺失信息

matlab中，fig图转pdf，缺失信息~直接把fig窗口缩小一下，再存储就ok了~

原创 latex 等号上面有字

$\overset{def}{=}$示意图：

原创 non-linear stationary subspace analysis with application to video classification

Stationary subspace analysis~Stationary Subspace Analysis (SSA)[1] is a blind source separation algorithm which factorizes a multivariate time series into stationary and non-stationary compo

原创 Image set based collaborative representation for face recognition

ISFR: Image set based face recognition难题：How to represent the query face image set by using the gallery face image setsSet-to-Set 的度量方法，忽略了训练集之间的联系~image-to-image的方法，忽略了set内部之间的联系~所以提出：A image

原创 Representation Videos using Mid-level discriminative patches + CVPR13

Video representation based on mid-level discriminative patchesMine these patches from training videosThen use these patch as a discriminative vovabulary for action classificationrepresent videos

原创 行列式的乘法

行列式detdet（A*B）=det（A）*det（B）；

原创 kernel learning for extrinsic classification of manifold features+CVPR13

Kernel selectionFeature lie on Riemannian manifold with known geometry.现在很多学习算法（如：判别式分析、partial least squares、和SVM)不直接应用这些特征，因为这些特征处在非欧式空间中。所以就面对这kernel选择的问题~这篇paper就是通过kernel learning的方法来选择合适

原创 矩阵logm和expm的定义

原创 Inductive hashing on Manifolds

Inductive Hashing on manifolds (Fumin shen, chunhua shen......)主要工作：How to learn compact binary embeddings on their intrinsic manifolds.（Non-parametric manifold learning）Learning based hashi