Jekyll2026-10-29T18:53:15+08:00https://tauq1u.github.io/my_blog/feed.xmlTauQ1u’s BlogPersonal blog of TauQ1u — Computer Science and Technology undergraduate at NCU, member of GOODLAB.TauQ1uResearch Notes 012026-10-29T00:00:00+08:002026-10-29T00:00:00+08:00https://tauq1u.github.io/my_blog/posts/Research-Notes-01How to find a good idea?

​ Recently, I got a task to rewrite an article. But before we decide how to approach the rewrite, we need to come up with an idea for adding some valuable highlights.

​ My senior labmate told me that I can boldly propose a new plan, and he would check whether the idea is worth exploring. However, I misunderstood what my senior meant. I naively thought I don’t need to read the paper deeply and that just using my brain to think was enough. Finally, I couldn’t find any useful methods to produce new work.

​ Three months later, I reread the paper and carefully examined the response letter. I found that the response letter had already pointed out the likely direction for making new contributions. Then, I shared my thoughts with my senior. Only then did I really get the project moving.

​ In my opion, if you need to come up with a good idea, you have two main ways. The first is to find out what questions the industry really cares about. Then, following these quesitons, you can try to find solutions to these problems. The second is to read the response letter to some high-level paper. However, we usually can’t get access to the response letter of a paper. So, we need to read the paper carefully. Many papers are the reflection of an author’s thoughts. In this case, it hard to avoid incorporating the author’s subjective thoughts. So, what we can do is to find defects of the system and made some innovations like “A is good, B is good, but A + B is going to be better”.

]]>TauQ1uResearch Notes 022026-10-29T00:00:00+08:002026-10-29T00:00:00+08:00https://tauq1u.github.io/my_blog/posts/Research-Notes-02How to reproduce the code

​ In scientific research, we usually need to reproduce the code. But how should we reproduce it? And do we need to simply complete the process, or obtain graphs similar to those in the paper?

​ One opinion I agree with says that we first need to run the code. This means we need to check the path, dependencies, and environment to ensure they fit our device. Then we need to print out the data flow direction. Then we can add comments to the code. This is a fast way to help us understand the code and quickly get started, but it won’t be comprehensive.

​ There are many opinions about how to improve out coding ability. But in summary, there is a common step: adding comments to the code. So it is obviously that we cannot avoid understanding the code. AI is a good way to get a result easily, but it is not healthy for our ability.

]]>TauQ1u