仅显示标题浏览

Bati's eHome of Tech

  :: 首页 :: 博问 :: 闪存 :: 新随笔 :: 联系 :: :: 管理 ::

How to do research?

  • Find a research point.
  • Read new paper,such as ICCV,CVPR,PAMI.If you find some new idea,which can be made use of in your work,you can do it.
  • Think it before you do it,analyze why you do it like that.
  • If you have a problem,talk about it with your teachers,and colleagues,analyze why it doesn't work.Brain storm...
  • Do not depend on others excessively,you should have your own ideas.
  • When you talk about with others,you should know what you can get from others.If you have any problem, must solve it.Do not leave it alone.
  • Study how to write a good paper,including structure,grammar,...,ensure readers can easily understand what you want to express, and your experiment support your idea.

A “Template” for Good Papers--From the report of Professor Charles Ling

  • Abstract What is the problem? What is your solution and results?
  • Introduction Problem X is important.Previous work A, B, … have been studied A, B have certain weakness (be careful here);We propose D, features of D (how much)Experiment with D.Compared with A, B,D is better than A, B(in certain aspects) ;Outline of the paper
  • Review of Previous Work Previous work:may split to several classes,e.g. Can review each work in one or several sentences;Compare to yours (refer to later sections)May also put it after sections about your work.
  • Our Work Definition, notation (need motivation);Algorithms: pseudo-code; diagram;explanations Answer potential questions from readers;Too much details (such as proof): appendix
  • Experiments and Comparisons Experiment design;Comparisons (Is it scientific?); Discussions (make sense of the results);Draw conclusions;
  • Conclusions Quick summary;Future work

 

zz from http://www.jdl.ac.cn/user/sfxie/Publication.html

posted on 2009-07-29 15:21  Bati  阅读(315)  评论(0编辑  收藏  举报