科学研究方法论

1. 做研究3W原则：
   what’s the issue?
   what’s new?
   Why’s important?
2. 提出问题是解决问题的一半
   辩论找到没有解决的问题
   科学是就大部分是猜
   解决问题黑道白道的人都要有
3. 实践是检验工作的唯一标准
   数学方法是为了解决实际问题，从实际问题出发，创造新的方法和理论
4. 从厚到薄 - 问题分类（复杂问题拆成简单问题的叠加）
   个人感觉：确如托福独立写作，将论点段的结构划分清楚，每一句写什么便很清楚
   中间的科学问题：分类的话有几个类？怎么分？
5. 杨振宁回答：物理有需多方向（其他学科也是），不是说一定需要数学才能做物理（MRI的例子：仅仅把已有的仪器中磁场从uniform改成non-uniform，卫平SVC的例子），但一个人研究科学需要找到自己感兴趣，并且擅长的方向（杨在国内西南联大有7年的场论的深刻学习，促成了后续的Yang-Mills理论）
6. 如何搜寻相关Topic的理论/资料(信息科学为例)
   搜关键字：百度谷歌/github/youtube/淘宝
   学校在线图书馆/经典教材

Miscellaneous 补充：
Tips I learn of doing research

1. Keep asking why and then find something that nobody knows. (Video Technology by Pro. Li)
2. You can never learn all papers or knowledge (there are tons of knowledge or paper being created a day), finding the most useful one efficiently. (Cloud Computing Network by Pro. Zhang).
3. It is not that finishing all the background knowledge you needed and then start doing research. You learn what exactly you needed. (Cloud Computing Network by Pro. Zhang).
4. Track specific scholar, not a research area. (by Pro.Zhang wenyi)
5. Educate your supervisor while being educated. (by Pro.Zhang wenyi)
6. Get mostly done yourself. (by Pro.Zhang wenyi)

From VideoTech class Pro. Li weiping
Before college, there is nothing in your brain, the purpose of studying is to learn knowledge. As a college student, you must learn the ability of finding knowledge and finally, the purpose of graduate course is not about teaching knowledge, is about showing you how to create knowledge. Ways: by continuously asking fundamental questions, find the answer and keep asking until you find out something that nobody knows before.

1. Sometimes when you find something new, you want to compare with previous, you should ask yourself, what is the limit of your findings, should you put more time on finding more.(Entropy coding, which one to ask question)
2. Double check your result, go deeper into fundamental questions (very beginning one), most significant results come from that. (Huffman coding, which one is last to ask, FGS)
3. Look at the whole picture, the fundamental thing, do not be misled. (Transformation coding, back in history, finding transformation transmit as less number as possible, but failed in compression)

From DSP Pro. Dai lirong, the process of doing research:

1. Problem introduction
2. Describe in math. (e.g. J=min[en]2)
3. Find method (e.g. Weiner- Hopf equation)
4. Method whether can be optimization using new algorithm(e.g. Recursive like LMS)
5. Analysis, finding the meaning behind the equation, which also is an important process.(e.g. Error signal of ADF output is orthogonal to input signal)