NOVEMBER

但是还是比不上压缩到一页的过程
因为好不容易搞出来的精彩句子都不得不删掉>_<

Statement of Purpose

I would love to pursue a Ph.D. degree in Department of Computer Science at University of Southern California, with a focus on the specialized field of computer vision.

Computer vision is magical. In this field, researchers often succeed in simulating the ability of biological vision systems to understand the visual world, and in coming up with systems working in real, noisy, and non-ideal situations, like football robots. This process would naturally require combined efforts from various disciplines. In fact, techniques adopted from fields such as machine learning and cognitive sciences have actively led to numerous progresses in computer vision, and are promising a more exciting future. However, the large amount of visual data vision systems have to process has been an unresolved fundamental challenge which, for example, has prevented face recognition systems to become fully applicable in reality.

Recognizing the challenges, I would love to devote myself to further studies in computer vision, and I am confident of being a competitive candidate. With 100 in Discrete Mathematics, 99 in Data Structures, 98 in Digital Image Processing, and 96 in Algorithm Design & Analysis, my undergraduate courses have given me strong mathematical and programming backgrounds, together with an initial interest for computer vision. Moreover, at the State Key Lab of CAD&CG, I have been taking training projects on computer vision since last year. They have introduced me to a variety of fundamental concepts and methods in this field. First, the stereo matching project demonstrated the power of energy minimization frameworks, for which I implemented two important optimization methods, namely, Belief Propagation and Graph Cuts. Later, while trying to write a KLT feature tracker, I read a wide range of papers on feature extraction, image filtering, and tracking. They equipped me with preliminary knowledge for a variety of tracking problems. Besides, by keeping track of recent progress in determining optical flow, I have got a clear understanding of the need for image statistics and machine learning techniques to infer structures from them.

The problem of image/video segmentation has gradually grabbed my attention along the way. Bridging between low-level details and higher-level semantics, segmentation results can offer significant help to solving other interesting problems, such as object recognition, which require object-level information. At USC Computer Vision Lab, the emphasis on segmented symbolic descriptions of objects matches my interests very well. What is more, given the wide outreach of research and the awareness of inter-disciplinarity here, I am sure that this is an ideal place for high-quality cooperation and the most cutting-edge achievements in computer vision. I plan to obtain my Ph.D. degree here, and make my own contributions to its research.

To pursue a Ph.D. is a precious opportunity for me. In the five years or so, I will try my best to train myself for computer science research by aiding creative activities at USC, continuously pushing myself to new limits. Yet I understand that obtaining Ph.D. is only the very first step in my academic career. After that, I want to go on with my work during Ph.D. creatively, and will be most joyful if my work could contribute to the progress of our computer science society, especially that of the vision community. Besides, I honestly hope that my work will provide value to real-world applications.

I am eagerly looking forward to being part of University of Southern California and the Computer Vision Lab, and I will Fight On!

 

Signature: Kun He             Date: November 24, 2009

放弃Gatech，虽然材料已经寄了。网申倒是没提交。
UCSB与UCSC，选择了UCSC。

现在的14所：
UIUC
USC
UT Austin
UMD
PSU
Brown
UCSD
UCSC
UPenn
UMass
Boston
SUNY-SB
UBC
Toronto

该稳定下来了……

这就是情况。从前一段时间的around 1:30AM 到这两天around 23:30PM，已经可以表扬一下了。
当PhD（至少是CS的PhD）之后又会往后推，与此图接近重合。这是必然的，不可抗拒的，也是合理的！！！！
可以说我做得不对，但是那么多PhD都大错特错了？单从我们lab看来，0:00左右是他们无可争议的黄金时间。难道没有理由吗？我想不出来比图中更好的两个理由。尤其是第一个！
忙一天，就指望着晚上能静下心来好好写东西啊。Deadline就在面前啊。你以为我不想多睡啊。


另，想考虑一下UCF。Vision组Marshall Tappen明确招人，对他的工作有了解，仔细研读过他03年的一篇一作文章。Orlando这个城市也还不错。但CS整体和学校整体都弱，学校最近相当缺钱，EE不平衡，周边IT相当薄弱。

Yeah.
I love USC.
And I will let USC know that.

----
And by the way, I've started to seriously consider UCSB and UCSC.
Funny isn't it, UCSB UCSC UCSD if I apply to them all.
UCSB: 2 labs related to vision. Beautiful. Owns a beach. Good Sci&Tech but others just suck.
UCSC: Wonderful location(very near silicon valley, beautiful scenery), CS pretty good, and a pretty good vision lab. But famous for rejecting people.

----
I may have played it a little bit wrongful: too early to publish.

UNIVERSITY	EE MS DEADLINE
UBC	12/01/2009
PSU	12/15/2009
UIUC	12/15/2009
UT Austin	12/15/2009
Gatech	01/01/2010
UCSD	01/04/2010
Toronto	01/07/2010
UMass	01/15/2010
SUNY-SB	01/15/2010
UMD	02/01/2010
Boston	04/01/2010
USC	04/30/2010
Brown	05/01/2010
UPenn	05/01/2010
RPI(?)	01/01/2010
SFU(?)	01/15/2010

已然进入电子邮件焦虑期了……

==USA==
UIUC
UMD
USC
UT Austin
GaTech
PSU
Brown
UCSD
UPenn
UMass
Boston
SUNY-SB
==CANADA==
UBC
Toronto
==HK==
CUHK

热烈祝贺villain T二战取得满意的成绩！
各种撒花！

HU12H@NG, you're such a dirty bitch!
I will let you swallow it. Just wait and see!
Cause now I'm not the one that fears.

越来越感觉到老鲍是好人……可惜不能达到理想的状况太多了啊啊啊啊啊啊啊>_<

UIUC
GaTech
UT Austin
UPenn
PSU
UMD
Brown
USC
one of (UCSD/UCSB)
SUNY-SB
BU
UBC

做决定总是最花时间的。

保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默保持沉默

合寄，合寄，考试，考试，网申……

啊哈哈哈哈～

今天你好像隐形了一样，我觉得你QQ上应该看到我了。不知道具体是什么情况，但是我可以猜。大约就是考试在即加上挂念手机压力大身体感觉不太好心情有点郁闷自己努力调整中吧……

我很抱歉，我正在经历非常忙的几天，周四之前不可能有好转，周四及之后也不见得会好转。

加油。没问题的。

如果所有学校的网申系统都做得像MIT一样简单直白，如果所有学校的要求都像MIT一样简单明了，那该多好！

……………………

紧张啊。

手机的事情有眉目了！真不知道那人怎么想的……

Seems that I'm more of a vision guy than a learning guy. Vision更偏工程，极端是Robotics；Learning更偏数学，极端是Statistics。明显都不是我喜欢的。更温和的也分两种，用learning去做vision，或者做learning然后用到vision中。现在看来我比较倾向于前一种，这跟我的经历绝对是分不开的，但是有可能也有想法上的原因。问题是过几年有没有可能往另一个方向走去？现在是难以否定的。龙哥就是一个活生生的例子，但我跟他的不同在于我没有他那样的firm belief，我甚至认为那样有点可笑。要说到learning，其实我更感兴趣的可能是human learning/perception，而不是现在这个样子的ML。阻力有如下：对所需的数学稍微有点恐惧；不太看好前景；不熟悉。张志华老师的讲座起了很大作用。如果是vision呢？问题摆在那里你总是要去解决，只是方法会进步。对了，这是我感觉比较舒服的想法。就这么着吧，结论：I'm still more of a vision guy than a learning guy.

要学会适可而止。

别说“旧的不去新的不来”。

好吧，这次逃过一劫，不过好运气不是每次都有。

南京没有让我失望，确实像我想象的一样有意思。

[The 23rd day]

莫过于下午5点半左右老鲍在我身边坐下，和颜悦色地跟我说了一会话。而且是在他骂完另一个学生之后！
不管怎么样，感觉很好，心里有点热乎乎的。貌似老鲍有点喜欢我了？
其实他本来就是一个很好的人，虽然表面上不像。
如果不是要出国，留在CAD或许也挺不错的。但是转念一想，学术上毕竟还是差很多啊……现在在CAD感受到了一些很好的气氛，但很有可能出国以后感受到的气氛又要好得多。
所以还是要像狗一样的挺过去，出国……

[The 23rd day]

Fuck!