Information-based Space

I spent these days in reviewing statistics-related stuff. Something like information theory, probability estimation etc. Many useful methods in PR or CV are based on information theory, like: HMM, MRFs, EM method, ICA, GMM and so on.
When googling, I found mutual information is also used for feature selection, which is called Minimum Redundancy Maximum Dependency method. Maybe it can be also use for feature extraction, like LDA, but it works in a "probability-based space, which adopts the mutual information or symmetric KL divergence as the metric.
In the next few days, hope to have time review something about affine geometry and functional analysis. I really can't afford to waste a sec. Of course, allways keep thinking is a good habit:-)

some new plans

It's hard to form a good habit, but it is worth. So I think something must be altered to do research efficiently.
1. Use time efficiently.
(Most important)
DO NOT spend much time on searching, reading and thinking is more important. So before the searching, make sure to know what you want  and keep a road map in case of being lost.
Abstain from the reading and talking about political topics. What you say makes no differences for this country, for you are not the saviour. Doing research is the best way to pursue your freedom.

2. Adjust the daily route. Get up a little earlier and sleep before 12. Good for both health and research. Try to study in the lib or other quiet classrooms. Think what you've done everyday when go to  bed. Write some research-related stuff in the blog when arriving at the dorm.

3.Keep thinking and writing down your ideas. Though most of them have been done or even perfectly by others, whatever, you can avoid wasting time on it. Try to contact with top researchers.

也谈沙叶新在港中文的演讲

粗略看了沙叶新在香港中文大学的研究后，他希望中国民主自由的热情让我感动，但是回想当年"百家争鸣"后的反*右，我不禁又感到悲凉：如今的GCD和昔日的有什么区别么？否也：
1.GCD始终没有停止对文化、思想的控制，而且冠以稳定压倒一切的口号。洗脑教育与当年比，有之过而无不及。这点从胡core上台来对网络的打压可见一斑。
2.GCD于心不诚，对外一套，对内一套，因为他们害怕让百姓看到国外报道的真相。而且他们还会动员自己控制的各种通信资源掩盖真相。这点可以从国内发生频繁却无媒体报道的暴*动看出。对待知识分子，他们的手段估计从来没有改变过。
3.GCD未表现出任何民主的意向。既得利益者不愿放弃自己手中的权利和资源，即使内部有些民主呼声，也会很快打压下去。

因此，温的讲话不可信，中国实现民主也很难指望他们。除此外，国内人民的开化也难能一日实现。
本人的期望：
能有似孙中山先生的人士出现；能有似蒋经国先生的leader(slim hope)；在保留中华文化的前提下从国外引进民主。

a good website for optical illusion

http://www.michaelbach.de/ot/sze_moon/index.html
This web collects many optical illustion examples. Some demos are very cool.  Now I am clear what I've seen may not as what it really is. I was cheated by my brain instead of my eyes, for the learning ability of the brain is far complicated than eyes. Hope one day I can get an idea to utilize these illusions

最近的流水账

早上去跑到光楼旁听现代微分几何，发现教室没有人，后来路过一MM，打听得知课从4月开始。然后在教室看自己下载的lecture notes，诺大的教室只有一个人，于是边看边听一下音乐。甚爽，发现一个好去处。

下午得知MSRA的师兄中了cvpr的oral，羡慕之余也激励了自己。上学期选了一大坨没用处的课程，这学期不能这样耗费时间了，提高自己的效率，要有自己大体的计划。

然后去图书馆自习，发现一个问题：再用google的时候，很容易偏离最初的目标。比如要查关于的A的资料，不过由于资料丰富，最后可能跑到Z上去了。效率不高。找到这样一个方法来解决：打开一个文本编辑软件做记录，比如今天查资料：Calibration->Camera Distortion->Brown-conrady model ->Fisheye lens distortion->Chromatic aberration. 大约4、5步左右，太长就返回。中间遇到问题或者有些想法就马上记下来，以后还可以留作参考。这样可以清晰的看到的自己的目的和当前的步骤。

 

还有个问题：本本用的是ipw3945无线网卡，ubuntu edgy下用network manager老会出现掉线，然后就开始拼命扫描网络，导致死机状。有哪位过路人知道如何解决，麻烦告知。PS：2.6.17 kernel generic.

张江之行

开学第一次到张江，可惜的是没吃上这里的早餐，因为上学期来了一次发现这里的早餐味道不错。不过今天又发现另外一个好处，WIFI网络速度奇快，而且可以联通国际网，research channel是相当地流畅，网速又几百K. 可惜来到本部的图书馆只能上国内网，国际网还要用宿舍的代理，速度也不快。 看来，必要时可以去张江自习。
谈到研究，继续修炼内功中.....