Data, Knowledge & Life

Weiwei Cheng's blog

Archive for 4月 2010

A desktop before the conference deadline [PIC]

leave a comment »

ECMLPKDD 2010 paper due: 30th of April 2010, midnight PT

Written by Weiwei

30/04/2010 at 21:03

发表在 杂话

The key word is learning, of course!

leave a comment »

It is the word cloud extracted from the titles of accepted papers at ICML 2010.
Click the image for an enlarged picture. You can also check it at Wordle.net.

Written by Weiwei

22/04/2010 at 13:28

发表在 学术

ICML 2010 Paper Acceptance Notification

leave a comment »

All of my three submissions, joint work with Krzysztof Dembczynski and Eyke Hüllermeier, to the 27th International Conference on Machine Learning (ICML 2010) have been accepted! These submissions were:

Label Ranking Methods based on the Plackett-Luce Model,

Graded Multilabel Classification: The Ordinal Case, and

Bayes Optimal Multilabel Classification via Probabilistic Classifier
Chains
.

ICML is the leading international machine learning conference, attracting annually some 500 participants from all over the world. ICML is supported by the International Machine Learning Society (IMLS). The selection process of ICML is very competitive. This year, only 152 of the 594 submitted papers were chosen to appear on the program. Each paper was handled by an Area Chair and received two reviews in Phase I. After the author response, the Area Chair hand-selected additional reviewers in Phase II until a confident recommendation for acceptance or rejection could be made. In addition to the reviews and the author response, many papers were discussed to clarify their assessment. All decision recommendations of the Area Chairs were reviewed by the Program Chairs, and often additional steps were taken to reach a final decision.

Drafts of our papers will soon be available at my university page.

Written by Weiwei

17/04/2010 at 13:12

发表在 学术

Google有多少服务器?

with 2 comments


via gizmodo

It’s said that 2% of all the servers in the world belong to Google.
Click the picture for a closer look.

Written by Weiwei

15/04/2010 at 21:59

发表在 转贴

Microsoft Academic Search

with one comment

  Microsoft Academic Search 微软学术搜索(Beta)是这两天我最经常把玩的网站。和Google Scholar比较起来,MAS的界面非常独到,更注重搜索排名的表现。虽然从它显示的文章引用数量上来看,MAS搜索结果的覆盖性尚不及Google Scholar,但它的用户体验的确是一流的。

  根据MAS的数据,机器学习与模式识别(Machine Learning & Pattern Recognition)领域的十大国际会议是:

  •   ICML – International Conference on Machine Learning
  •   UAI – Uncertainty in Artificial Intelligence
  •   NIPS – Neural Information Processing Systems
  •   COLT – Computational Learning Theory
  •   EuroCOLT – European Conference on Computational Learning Theory
  •   ECML – European Conference on Machine Learning
  •   ICPR – International Conference on Pattern Recognition
  •   ICDAR – International Conference on Document Analysis and Recognition
  •   ICA – Independent Component Analysis
  •   ICVGIP – Indian Conference on Computer Vision, Graphics & Image Processing

  此学科的十大期刊是:

  •   ML  – Machine Learning
  •   NECO – Neural Computation
  •   IEEE Transactions on Neural Networks
  •   JMLR – Journal of Machine Learning Research
  •   JASSS – The Journal of Artificial Societies and Social Simulation
  •   PR – Pattern Recognition
  •   Finance and Stochastics
  •   Information Research
  •   PRL – Pattern Recognition Letters
  •   Internet Mathematics

  机器学习的同行们,这个排名里是不是也有你最喜爱的会议和期刊?看完这个排名,估计已经有朋友开始皱眉头。我在Twitter上的好友就有跟我说,NIPS怎么可能这么靠后?还有那个那个啥,怎么不在名单里?其实这没有什么太奇怪的。我猜测MAS的这个列表完全是依照文章发表数以及文章引用数加权算出来的,有些偏差在情理之中。

  MAS还提供了界面非常友善的学者搜索与排名。比如在机器学习与模式识别领域排在第一的就是传说中的Vadimir Vapnik。

  MAS为部分知名学者提供了相对完善的数据索引,包括最近比较流行的G-IndexH-Index。传奇的Vapnik有着H-Index值37。鉴于MAS一共只索引了他119篇文章/书籍,这个数字已经是不可想象之高了。高山抑止,膜拜中…… (It is a moment that you are NOT supposed to ask about mine.)

  总之,作为一款beta产品,Microsoft Academic Search已经颇为优秀。大家不妨尝试一下。

Written by Weiwei

12/04/2010 at 22:51

发表在 学术

谷歌服务在中国大陆的可用性

leave a comment »

googleinchina by you.

  出去旅游拍回来的照片我一般都会上传到Google Picasa相册上,留做备份,和大家分享。可惜国内的亲戚朋友访问Picasa的时候却经常出现问题,不是连接速度极慢就是索性无法打开页面。中国的海外华人,像我这样大量使用Google服务的应该不在少数。鉴于Google各项服务的可访问性在国内时常间歇变化,能够及时知道他们可否被实时使用是很有意义的,这也就是为什么Google提供了谷歌服务在中国大陆可用性的每日报告。

Written by Weiwei

08/04/2010 at 08:23

发表在 杂话