Creating mind map for research
·So till now I was trying to act out the research and wanted to decode what the research is through activities. Now after one year and reading so many papers. Somewhat a mind map is building in my head.
For me research constitutes two parts. first is research itself and second part is technology. In research we want to dig into something deep enough that we can uncover all layers. I always used to think that literature review is a thing via which we define how many directions we have explored but it is not. It is a part of writing only defining which part of the research is relevant to the reader. But research of a subject is to keep digging. But to what extent. First until you think that you have covered all aspects of it. And second you yourself becoming aware of the direction in which you want to explore the topic further. Is research about finding a solution? No. But basing your experiments on some theory or hypothesis and extending the human knowledge on the basis you chose. When you cant find any literature related to your direction, you design the experiment. Then secondly comes the technology part. When you pick up a direction, a problem, a question or a hypothesis to explore what is your approach to experimenting. I am still exploring this part. I tried to relate to different approaches people adopted. Like balanced machine learning, learning representation but these are all directions you go to for exploration when you actually have a direction or problem in hand. I have a lot of examples where attempt to find a solution was made. Like in thesis of kranti parida or in the papers of rishabh parihar. I will keep them in reference to follow. The most critical paper that I would like to follow is the paper from Rachavarapu on pseudo labeling. Thats me writing and trying to learn what is the technology or say technical part of the research which is mostly missing in the thesis of IIT Roorkee.
One more thing I want to keep reminding myself is that instead of running behind a solution, its more interesting to find a solution for a problem. I am learning how to approach a problem in research and how to find solution to it. Finding a problem is itself a wholesome task. Following it through to research and finding a solution is what constitute the above mentioned two research aspects.
Approach to a problem to research :
1) how one approach is carried out till now in combination with another approach
2) how the combination of these approaches is beneficial for the problem
3) doing experiments with them
Keep iterating these steps even in parallel and keep the speed of failure at max. Thats what research is. This is a slow process but I think it should work. Basic fundamentals still apply :
1) Building foundational knowledge
2) Pattern recognition in problem and solution
3) Thinking strategically in the process.
And above all keep asking yourself why…why I approaching the problem the way I am.