Forum:Can We Have A Collection Of Interesting Research Topics?

Question

[Michael]

I have seen these questions come up many times, they are were all phrased like: Please suggest interesting research topics for my masters, bachelor, or Phd thesis. Most of the time these questions didn't really contain minimal information about the specific interests, skills, and environment of the candidate, and so they were mostly closed, sometimes with comments, that suggest to be more inventive and show initiative on their own. I haven't seen a lot of really good project suggestions.

While I understand this sentiment, I still believe that the questions showed a requirement for guidance in choosing a topic. On the other side, there might be new supervisors in need for some ideas, good topics and advice for designing a good project for students. What is the point for an online community in this? I certainly see concerns

• We cannot and should not try to tell you what you want, because you should develop your own interests and show initiative
• We don't know enough about you and your research environment, or you wouldn't want to tell us enough about your situation in public
• A thesis needs to be tailored specifically to you and most importantly to your supervisor
• A thesis needs to be unique, if we post very concrete topics people might tend to re-use them worldwide
• You want to keep your most innovative ideas to yourself.

However I imagine, guidelines for choosing projects and topics will be very useful. Therefore, I would invite you to share your thoughts on how to search and select a good project, and also post more or less concrete project proposals, or where to find them (e.g. Google code,etc?). And also how to determine, what is an appropriate "size" for your project work.

[Michael]

Interestingly, nobody ever posted any project idea. I take this a proof.

Answer 1

[Michael]

I'll start with some general advice:

The most important persons involved in your thesis are of course you and your supervisor(s). You absolutely must talk to your supervisor and agree with her or him on the topic and the goals to achieve. It really helps to have a project the supervisor is interested in. Look around at your faculty, and talk to several research groups. Often, they will have some more or less defined topics on offer already.

Decide for a topic that really interests you, try to find a nice group which does very interesting research, and where you could imagine to take part in exciting projects. Maybe the group holds opportunities for after your thesis is finished? The research environment is actually very important. In bioinformatics, a lot of work is done in cooperation. In an optimal world there are researchers around to work together on a topic, you will have a lot of support and can ask questions and discuss your work.

It is also about your motivation and the skills you have. Try to summarize your skills, do you like programming, or work with databases, ...? There are different types of research projects, e.g. some that involve hardcore programming, others statistics, biological data analysis, evaluation of existing methods, or developing and improving algorithms. Avoid wanting to solve too many problems at once, focus on some specific and well defined tasks. This is especially important when it comes to thesis with tight schedules, you need to be aware about the timing of your project. Try to be realistic about what you can achieve in the time frame, it could be advisable to define tasks and then mark some as optional form the beginning.

"Is the data available?" should be another core question. This is such a crucial point in data-driven projects. Is the data you require for your work already available, or are you just given promises it will come soon? What if these are just empty promises? Can you generate data yourself, or use existing published data sets? Bringing existing data into shape can also pose some unforeseen challenges.

Once you have decided for a project and have an idea about what your thesis should be, write up a proposal and project plan, and have your supervisors agree on it finally.

Here is an interesting paper on choosing a topic.

Answer 2

[5heikki]

I would love to see a fully parallelized affinity propagation implementation or at least something that is better than what we have now (from what I can tell Concurrent_AP is currently the fastest AP implementation for larger datasets, especially so after you replace loadtxt with read_csv). There are papers which gloat about having made something better but no code is provided so yeah.. Feel free to implement :)