杰 - C++博客

combntns函数所引起的内存不足问题

编写NFL和NFP分类器应用于ASLAN数据库发现combntns函数所引起的内存不足，到网上搜索有人提问，但没有解决方案。
问题：我想取一个序列的任意组合，用combntns，在数目较小时，可以达到目的，但是数目变大时，就不行了，求助各位。

例如取1:5,里面3个数字的组合，很容易得到。
combntns(1:5,3) 就可以了

但是取1:100里面10个数字的组合，就不行了。
我的解决方案：如果你未必要取所有组合，可以从1：100中任取20个数，再取这20个数字里面10个数字的组合。
%matlab code:

rand('state',0);

temp1 = randperm(100);

temp2 = temp1(1:20);

combntns(temp2,10)

posted @ 2012-12-13 08:30 杰哥阅读(818) | 评论 (0) | 编辑收藏

win7下matlab7.0的卸载问题,为什么matlab2010a只能在断网时才能打开？

http://www.ilovematlab.cn/thread-68860-1-1.html

在win7，32位家庭版下,有两种方法解决正常运行问题，一是将配色方案改为经典(在桌面空白处右击鼠标，选择个性化，在弹出的对话框中选择windows经典主题即可)，二是兼容性改为vista sp1或sp2.卸载完了后，再把主题改回到你原来的主题。

为什么matlab2010a只能在断网时才能打开？
20121207发现matlab打不开，上网搜索解决方案。有人说断网就可以。我实验两次，注销后断网能打开matlab，注销后连网，不打开其他任何程序，matlab还是打不开，等了五分钟。能看到matlab进程。继续搜索“Windows 7 matlab断网才能打开”。查到解决方案(http://wenwen.soso.com/z/q294348187.htm)：你可能用的是window7吧，这种现象的产生是由于MatLab与Windows 7 的兼容性问题，解决的方法如下：从纯净的Windows XP系统中system32目录中拷贝一份iphlpapi.dll到 Matlab 2010a安装目录\bin\win32中问题即可解决，当然，X64的系统要到 Windows XP X64 中去拷。我问Lianyang Ma要了这个文件，重启下电脑就搞好了(即使在同时打开qq,gtalk,谷歌浏览器,FileZilla的情况下，也能打开matlab)。与配色方案和兼容性没有关系。

posted @ 2012-12-05 19:32 杰哥阅读(1446) | 评论 (0) | 编辑收藏

全国接听免费附加套餐

http://service.ah.10086.cn/pub-page/busi/commonBusi/optpage_common.html?f=200011716&tfCode=JIETINGF&kind=200011524&area=cd

http://www.10086.cn/ah/，网上营业厅，不是点击“业务办理”，而是点击“换套餐”，长途漫游类，全国接听免费附加套餐

posted @ 2012-11-23 17:46 杰哥阅读(1472) | 评论 (5) | 编辑收藏

arXiv

地球上首屈一指的科学论文预印本网站，科学出版业开放获取运动的领头羊，目前主站点落脚于康乃尔大学，在世界各地设有17个镜像站点，中国的镜像站点是 http://cn.arXiv.org 。

现今的数学家及科学家习惯将论文先上传至 arXiv.org ，再提交给专业的学术期刊。尽管 arXiv 上的文章未经同行评审，2004年起采行了一套“认可”系统。

arXiv.org的创始人物理学家Paul Ginsparg因之获得了2002年的麦克阿瑟奖。在其上能搜索Professor Tao的姓名就可以搜索到其论文。

posted @ 2012-11-23 16:31 杰哥阅读(2208) | 评论 (0) | 编辑收藏

Multiple kernel learning (MKL)

看论文SimpleMKL第一节标注部分就能完全掌握其概念

posted @ 2012-11-22 18:27 杰哥阅读(1083) | 评论 (0) | 编辑收藏

Subgradient method

http://en.wikipedia.org/wiki/Subgradient_method
Classical subgradient rules

Let $f:\mathbb{R}^n \to \mathbb{R}$ be a convex function with domain $\mathbb{R}^n$ . A classical subgradient method iterates

$x^{(k+1)} = x^{(k)} - \alpha_k g^{(k)} \$

where $g^{(k)}$ denotes a subgradient of $f \$ at $x^{(k)} \$ . If $f \$ is differentiable, then its only subgradient is the gradient vector $\nabla f$ itself. It may happen that $-g^{(k)}$ is not a descent direction for $f \$ at $x^{(k)}$ . We therefore maintain a list $f_{\rm{best}} \$ that keeps track of the lowest objective function value found so far, i.e.

$f_{\rm{best}}^{(k)} = \min\{f_{\rm{best}}^{(k-1)} , f(x^{(k)}) \}.$
下图来自: http://www.stanford.edu/class/ee364b/notes/subgradients_notes.pdf
例2：SVM代价函数是hinge loss，在(1,0)除导数不存在，取1和1之间的数值，具体怎么取？Mingming Gong said好像这个pdf和http://www.stanford.edu/class/ee364b/lectures/subgrad_method_slides.pdf，其中一个讲了。Mingming Gong asked tianyi, which is better, subgradient or smooth apprpximation?结论是不一定，subgradient解的是原问题，smooth不是解的原问题。一个相当于对梯度的近似，一个是对函数的近似，很难说哪个好。

posted @ 2012-11-22 16:08 杰哥阅读(842) | 评论 (0) | 编辑收藏

How to use matlab solve optimization quadratic?

Nannan gives me a fold named "Matlab Help". On page 46 of "Optimization Toolbox User Guide", it lists the constrain and objective type, and the matlab function. For example, if the constrain is linear and the objective is quadratic, we can use quadprog. Note that it can not slove $D_1$ in Section 4.1 of "Smooth minimization of non-smooth functions". Problem: max ((X^T)HX) and H is positive semi definite. The matlab function "quadratic" can not solve this kind of problem. It can only solve the problem: min ((X^T)HX) and H is positive semi definite.

posted @ 2012-11-21 18:31 杰哥阅读(672) | 评论 (0) | 编辑收藏

Cauchy–Schwarz inequality

http://en.wikipedia.org/wiki/Cauchy%E2%80%93Schwarz_inequality

The Cauchy–Schwarz inequality states that for all vectors x and y of an inner product space it is true that

$|\langle x,y\rangle| ^2 \leq \langle x,x\rangle \cdot \langle y,y\rangle,$

where $\langle\cdot,\cdot\rangle$ is the inner product. Equivalently, by taking the square root of both sides, and referring to the norms of the vectors, the inequality is written as

$|\langle x,y\rangle| \leq \|x\| \cdot \|y\|.\,$

Notable special cases

[edit]Rⁿ

In Euclidean space Rⁿ with the standard inner product, the Cauchy–Schwarz inequality is

$\left(\sum_{i=1}^n x_i y_i\right)^2\leq \left(\sum_{i=1}^n x_i^2\right) \left(\sum_{i=1}^n y_i^2\right).$

posted @ 2012-11-21 11:03 杰哥阅读(489) | 评论 (0) | 编辑收藏

Lipschitz continuity

Given two metric spaces (X, d_X) and (Y, d_Y), where d_X denotes the metric on the set X and d_Y is the metric on set Y (for example, Y might be the set of real numbers R with the metric d_Y(x, y) = |x − y|, and X might be a subset of R), a function

$\displaystyle f: X \to Y$

is called Lipschitz continuous if there exists a real constant K ≥ 0 such that, for all x₁ and x₂ in X,

$d_Y(f(x_1), f(x_2)) \le K d_X(x_1, x_2).$ ^[1]

Any such K is referred to as a Lipschitz constant for the function ƒ.
Reference：http://en.wikipedia.org/wiki/Lipschitz_continuity
区分概念，Lipschitz Continuously Differentiable，见Nesterov’s Optimal Gradient Method(Yaoliang Yu的ppt第十页)

[zz] 关于Lipschitz连续性的几个例子 http://blog.sina.com.cn/s/blog_544a70700100fqix.html
1，f(x) = |x|

是Lipschitz连续的，Lipschitz常数为1，但是其不可微

2，f(x) = x²

是Lipschitz不连续的，在无穷大处，f'(x)=2x,趋于无穷陡

3，f(x) = √x defined on [0, 1] is not Lipschitz continuous. This function becomes infinitely steep as x approaches 0 since its derivative becomes infinite.

4,The function f(x) = x^3/2sin(1/x) (x ≠ 0) and f(0) = 0, restricted on [0, 1], gives an example of a function that is differentiable on a compact set while not locally Lipschitz because its derivative function is not bounded.

f'(x)=3/2√x sin(1/x)-cos(1/x)/√x

趋向于0的时候无界

posted @ 2012-11-20 10:42 杰哥阅读(1419) | 评论 (0) | 编辑收藏

The difference between Fisher’s Linear Discriminant and Linear Discriminant Analysis

Fisher’s Linear Discriminant is for classification while Linear Discriminant Analysis is for dimension reduction. How is Fisher’s Linear Discriminant used for classification? The idea is very simple. It first reduces the dimension to one dimensional space using Linear Discriminant Analysis and then choose average of the two projected means as a threshold for classification. Therefore, the code is very simple to implement. See pages 13-18 of the slides "Linear Classification" of my computer for details.
The differences between them have been mentioned by Professor Tao.

Other reference: "Introduction to Machine Learning-dietterich" of my computer

posted @ 2012-11-14 12:05 杰哥阅读(595) | 评论 (3) | 编辑收藏

常用链接

留言簿(57)

随笔分类

随笔档案

相册

Other

Paper submission

福彩

留学相关

论坛

搜索

学者

邮箱

中科大和中科院

搜索

最新评论

阅读排行榜

评论排行榜

http://en.wikipedia.org/wiki/Subgradient_method
Classical subgradient rules

Notable special cases

[edit]Rⁿ

常用链接

留言簿(57)

随笔分类

随笔档案

相册

Other

Paper submission

福彩

留学相关

论坛

搜索

学者

邮箱

中科大和中科院

搜索

最新评论

阅读排行榜

评论排行榜

http://en.wikipedia.org/wiki/Subgradient_methodClassical subgradient rules

Notable special cases

[edit]Rn

http://en.wikipedia.org/wiki/Subgradient_method
Classical subgradient rules

[edit]Rⁿ