TL;DR. You have to be passionate and insane. I would not recommend doing it.
Research is an exciting thing. The purpose is to produce knowledge. You have the ultimate freedom because the job is about extending knowledge towards unknown.
1. Love Computer Science
You have to believe in what you are doing.
2. Learning Programming
Programming is a must. You have to write the implementation of your idea. And don’t forget the annoying part! Data processing will take the majority of your time, and there is no open source solution for that.
In some cases, you will need to use a lot of languages. I used C++/Java/Python/Ruby/Bash. Just learn.
3. Learning Mathematics
Things are useful: Linear Algebra, Probability, and Calculus. It’s for you. You don’t want to cry if you can’t even guess what theory belongs to from the paper. Knowing that you have to learn if you need something otherwise, you have to give up for your own thoughts.
4. Find good peers
Unless you have equipment and money. But wait a minute, it’s the least problem. If you don’t have a supervisor, you will have no way of knowing if you are actually thinking. Generally, more experience is better.
5. Find awesome papers
You are not alone. Papers are written by other miserable researchers. They tried to write it as well as possible. Welcome to this community. But you need to find good papers to start. Learn Scientific Method. If you failed on this, you can’t define your research problems.
6. Learn everything you need
Let’s read this paper in a day. Sorry, I mean it will usually be a week if you are not familiar with all the terms. Machine learning or deep learning, it doesn’t matter. Learn.
7. Be passionate and have some fun
Research is a lonely job. It’s one of the hardest jobs. Nobody can help you when you are creating. It requires hard work. And good motivation is from your position. And You need to rest. You need to get some good feelings from creation. Your brain is just not that good when it.
8. Believe it just takes time. A lot of time.
Learning takes time depending on how smart you are. I’m really dumb at getting all of that new knowledge. Programming takes time depending on how annoying the problem is. And your trials are part of the game. That takes weeks until you build a new mental schema.
9. Design your strategy and scientific problem
You need to focus on a tiny piece so you can work on even your research topic is fantastic. Keep looking until you have a good one.
10. Accept your ideas are shit
The secret is that it will get better if you think about it. Don’t stop.
11. Make a time plan
It’s not about catching the deadline. It’s for stopping. Can’t you do your research perfectly right? You have less data, limited time, time schedule, your future, and your health because you are exploring alone.
12. Treat people nicely
There are many researchers in the world working on the same thing. Write something that’s easier to understand so that you can also read a good paper. Reply to inquiries so that you can get a reply from the community.
13. Find more people
You can’t work alone when it’s starting to emerge with new directions and new ideas.