-1

I have recently (a couple of months) started a Ph.D. in Germany in the field of reinforcement learning, which I try to combine with Optimal Control to give guarantees and /or analyze performance and convergence properties and the like, or at least that was the initial idea.

The thing is, I am not enjoying the experience so far. I was never interested or very good with theory and math proving, but I was applying to many machine learning Ph.D. programs last year, and that was the only one I got accepted in (or at least the first). Because the salary is tolerable and the overall program prestigious, I thought I might as well take the risk and plunge in and push myself out of my comfort zone, even though I was, and still am, very uncertain I would succeed.

I do not like it so far for several reasons, which I will provide for context. First, the lab itself: While the other Ph.D. students are likable people, and I have no issue with them, they are all essentially doing theoretical research in control theory, and one tries to combine it with deep learning. Besides that, there is no machine learning or even much data-driven work. Ergo, I am in a lab where I neither understand nor care about anyone's research, which feels isolating and, quite frankly, useless. There's also the issue of teaching assistance, which I have been told takes substantial time ("We don't have time to do research over an entire semester" time). I can experience it for myself already as I am supposed to be a TA at a theoretical optimization course with material I don't understand.

Then it is the issue of the supervisor. The lab's PI is a super senior Prof with tens of thousands of citations and over 10 Ph.D. students. Mine is a super junior Professor who just started on the tenure track a little over six months now and has two other Ph.D. students. I feel our mindsets don't match my Prof's because he's way too theoretical. At the same time, I am much more software-oriented and even feel he is trying to micromanage me by telling me how many hours I should spend on X course, always letting him know what I am working on, etc.

He is even opposed to the idea of me doing projects outside of work, like an entrepreneurship incubator, because "before I have results, this kind of work, as well as networking, are useless," which I honestly find idiotic. He even lectured me when I let him know I would join, even though I had asked him before applying because he said I should listen to him more and tell him these things like I would tell my friends and some other nonsense. While the program will last around six weeks and require my presence for six days, I still do not think that reaction was warranted or useful. Truth be said, I dislike him increasingly as time passes, and I am afraid I will hate him in a few months.

Finally, there is the issue of the work itself. I just don't find myself excited to try to make sense of papers all day. It feels dull and meaningless. I am still figuring out my exact research topic too. Still, I am thinking of going into more causality-based learning, which I am surprised the Prof sort of accepted after several attempts because while he constantly says he is flexible, the first few times I suggested some topics beyond the very initial rough ideas we had sort of agreed on, that is the theoretical analysis of RL algorithms, he has several times said he is more interested in staying the course on those initial ideas and not exploring more, which feels constricting. I mean, the whole point of a Ph.D. is to explore relevant topics freely, isn't it? Right now, I feel less free here than when working in the industry.

Therefore, and because I do not believe those problems can be overcome, or even that I want to overcome them, I have come up with the following plan: I will give this program some time so that I am confident beyond reasonable belief about these observations and to figure out I indeed do not enjoy or am good at this kind of thing. I want to clear any doubts I may still have while trying very hard to make it work and not have any thoughts about quitting without a fight.

I will minimize my time toiling on my Ph.D., working diligently during my 9-5 but no more. Instead, I will dedicate significant time to upgrading and expanding my data science and software engineering skills and preparing for technical interviews. Then, if in one year the situation has not dramatically improved, I will apply for jobs in the tech area in the US and quit the program after I have secured an offer. If I am asked by employers why I quit the Ph.D., I will sell it as the sort of thing I just realized is not for me, and I want to create value, work with code, etc.

What does the community here think about my thoughts and plan? Is there anything I need to improve or rethink?

2
  • 4
    This site is not intended for questions of the form "here is my situation, any comments?". That said, you absolutely shouldn't do a Ph.D. in a topic you are not interested in, especially if you don't relate to the research that the rest of the group is doing either. Moreover, if you find reading other people's papers dull and meaningless, it's not clear that doing any Ph.D. at all in CS is a good idea. Finally, it's not true that "the whole point of a Ph.D. is to explore relevant topics freely", and it is reasonable for the advisor to put some constraints on the direction of your research. Jun 4, 2023 at 11:58
  • 1
    This is not the right forum for this type of "question". But for what it's worth, reading your complaints makes me think you're way too far away from where you'd need to be to succeed in your current position. Your goals and judgements seem barely connected to those in your environment. In fact they seem actual opposites. Don't try to fit a square peg through a round hole! Jun 4, 2023 at 16:00

2 Answers 2

3

We can only go with what you wrote here, and given that: if you do seriously want to give it another year to decide whether this is for you, then you do need to reflect on and rethink how you view things.

If you are not interested in what anyone else is doing, if after six months you qualify suggestions from your supervisor as idiotic and if you deem a core part of academic research (reading papers) as dull AND if you are convinced that a PhD means unlimited freedom for a PhD student, then you are likely fully unaligned with your PI and your surroundings. It sounds like neither the lab nor the topic nor perhaps a PhD in general is what you want.

Bring this up with your PI as soon as possible. They will want to know your perspective and - especially if they are on a tenure track- there is as at least as much in it for them to make your PhD a success as it is for you. If you cannot come to a better understanding of your respective expectations and responsibilities then don’t prolong your PhD - this is not failure on anyone’s end, sometimes things simply don’t work out and it is in everyone’s interest in the long run to acknowledge that - as hard as that may be and as painful as some of the conversations you are going. to have may be.

0

Finally, there is the issue of the work itself. I just don't find myself excited to try to make sense of papers all day. It feels dull and meaningless. I am still figuring out my exact research topic too.

Voilà!! You unearthed underlying issue.

For a PhD, give this careful thought and work around and towards it.
Obviously, it's a continual learning/discovery.

In research, you need to make sense of (relevant) papers. Find a strategy that works.
You may check out

There are tons out there.
Whatever you do, if you desire interest continuing your PhD, don't get to the point of hating your supervisor. You might just get self blocked and wander off ...

Not the answer you're looking for? Browse other questions tagged .