吴怀宇_中国科学院分享 http://blog.sciencenet.cn/u/wuhuaiyu 博士、副教授 「模式识别国家重点实验室」&「中国-欧洲信息,自动化与应用数学联合实验室」

博文

[CV论文读讲] The conjugate gradient method

已有 2316 次阅读 2012-11-30 14:18 |系统分类:科研笔记| 论文, method, gradient

Gradient descent method
Search line : gradient direction
The minimization in the search line must be  tangent to the contour of  f(x).
Problem: convergence is slow in most situations

So, using conjugate direction, for 2-dimension space,
we can find the minimization in 2 steps:
Find an initial direction, then search the minimization in this direction.
Find the conjugate direction, then search the minimization in the conjugate direction.

相关下载详见 “视觉计算研究论坛”「SIGVC BBS」:http://www.sigvc.org/bbs/thread-34-1-2.html


 



https://wap.sciencenet.cn/blog-4099-637812.html

上一篇:[CV论文读讲] PR08-马氏距离学习
下一篇:[CV论文读讲] Feature selection and extraction——mainly PCA
收藏 IP: 222.129.50.*| 热度|

0

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...
扫一扫,分享此博文

Archiver|手机版|科学网 ( 京ICP备07017567号-12 )

GMT+8, 2024-4-19 17:13

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部