4

I am doing my PhD in data mining / artificial intelligence and am at the beginning of my second year. I have had three master students so far, whose performance varies quite a bit. The worst (he already finished) got a very clear, cut and dry task for his thesis and still did not mange to do it as expected. The best so far (close to the end of his thesis) had a very vague task, and still got pretty nice results. The third is in the middle of his thesis.

In each each case, the really creative ideas, the innovation if you will, always came from me. Currently I have this third student in the middle of his thesis who has quite a few ideas how to use existing techniques, but I get the impression he is not really able to invent his own. Possibly I am not creating a good environment?

My students seem to lack quite a bit in technical know-how (pragmatic aspects of programming), mathematical understanding and knowledge of existing machine-learning methods. This is understandable – the end of your master is by far not the end of learning and I am happy to help them by explaining those things to them. Still, I struggle a little with that, since this is a road blocker. This is just learning, and not yet creating.

Now I have new task that I would like to have solved and which would make a fine master-thesis topic. However, no solution exists for the problem yet and it is pretty challenging, requiring quite a bit of creativity. A more or less completely new algorithm is required.

My questions are:

  1. Is there a point giving this task to a student or should I forget it and solve it myself?
  2. More general: What creativity can I expect from master students to create new methods? Am I expecting too much?
  3. How can I foster creativity in my students?
  • I think you're definitely expecting too much from the average master's student. – Massimo Ortolano Jan 13 '18 at 15:11
  • @MassimoOrtolano: Well, I did something (more or less) completely new in my master thesis and my brother is as least as smart as I am. Though my understanding has grown, I am not sure my creativity as such has. So I would have hoped there are some students as capable. So what do you think regarding my questions 1 and 3? – Make42 Jan 13 '18 at 16:17
  • In fact, you've undertaken a PhD afterward. But don't think just in terms of "smartness", there are many qualities that come into play in a successful work. For point 1, I think that you can give the task to a well motivated student if you know that you can provide some guidance toward the solution (if you don't have a clue, don't do it), and if you have a backup plan that would allow the student to graduate in a reasonable time. Maybe later on I'll try to expand everything in an answer if I can find the time. – Massimo Ortolano Jan 13 '18 at 16:39
  • Can you specify the timeframe of such a thesis at your institution and whether the students can exclusively devote their time to the thesis. – Wrzlprmft Jan 13 '18 at 16:59
  • You've been supervising a trio of students who're just slightly junior to you? How'd this come about? – Nat Jan 13 '18 at 17:27
2

I categorise potential thesis topics into high-stake and low-stake ones:

  • For low-stake topics, I have a clear plan what should be done in which order – unless some unexpected obstacle occurs or something new and exciting is found. The student still needs to apply their skills and knowledge to solve smaller problems along the way, but I do not expect any obstacles that I cannot solve with a day’s thinking myself. Thus, I can always keep the thesis work from stalling due to this reason.

    Such a thesis may produce a publishable result, usually if the student works efficiently and thoroughly.

  • High-stake topics (like developing a new method) involve a considerable amount of trial and error or a considerable creative leap. In particular, I do not have a good solution to the problem in mind (otherwise it wouldn’t be a high-stake topic). However, I have at least a few approaches in mind to ensure that the student has sufficient work to do.

    It may happen that the student will spend their entire thesis pursuing dead ends and their thesis will be a catalogue of what they tried and what didn’t work. The student may still get the best grade for their thesis, but it may be a very frustrating experience, and there is likely no further useful by-product from this thesis, such as a publication. On the other hand, if the student succeeds, this usually means that their work is really useful or interesting and likely to be publishable.

    I would never give such a topic to a student without informing them about the above risk (there always are some low-stake topics lying around; so they always have a choice). Also, I would only offer such a topic to a student who shows at least some promise in terms of enthusiasm, discipline, history, knowledge, and intelligence (though it is often not possible to assess all of these factors).

Note that, while there technically is a grey area between the two, few thesis topics actually fall into it (at least in my field).

Now, why am I telling you all this? If you want to provoke your students’ creativity, you need to give them high-stake topics that force them to brood over it for some time. However, this does not work for every student, and students should not be exposed to such topics only with prior warning. The latter is not only to allow them make an informed choice but also to prepare them for what is to come, should they take the challenge.

A few further notes and answer to your specific questions:

  • Is there a point giving this task to a student or should I forget it and solve it myself?

    Never rely on a thesis to produce anything. Even promising students may turn out to be overconfident, lack discipline or stamina. There simply is no feasible way to judge a student’s prospects for a thesis. If this is important or promising, do it yourself.

  • More general: What creativity can I expect from master students to create new methods? Am I expecting too much?

    That’s hard to answer as it depends on what makes for a new method in your field. In general the best I would expect is combining existing and adapting existing methods, not developing something new from scratch.

  • How can I foster creativity in my students?

    In my experience, to be creative about a given scientific problem requires:

    • Familiarity – give your students time and encourage them to familiarise themselves with the problem, investigate why established methods fail, etc. At least in my field, this makes for a considerable portion of the thesis time. Just writing down the problem and thinking very rarely works.

    • Inspiration – talk with them about the problem frequently; ask them questions about their approaches; encourage them to browse the literature.

    • Being open to failure – encourage your student to practice a lot of trial and error, and be clear that their is no shame in exploring dead ends.

    • Freedom and time – do not demand regular results from your students in a phase where they may need to reflect their results, failures, the literature, etc. Make it clear to your students that their problems require such phases.

0

Some people are capable of coming up with new algorithms and ideas, and some people just won't be able to do it. It doesn't mean that they're not capable of being good researchers—it's just that developing a new method from scratch isn't what they're good at.

I don't think there's a way to "make" students creative in algorithm and method development. You can try to "screen" for it—give them a task and see what they do to solve it—but trying to force a student who isn't skilled enough to do method development to do so is asking for trouble.

  • In my research group the task of us researchers is to develop new methods. Of course this also entails a lot of grind work like making experiments on example datasets (and especially finding useful datasets for demonstration), but the goal of the publications is indeed to present new methods. In my job in industry I am also not able to always just pull algorithms from the shelf, but need to develop my own ideas. – Make42 Jan 13 '18 at 20:17
  • So while I agree that not everyone is cut out to develop new methods, I have to ask what (mathematical) engineers do our universities create then? For my masters I study electrical engineering. I feel, in a way engineers are people "building cool stuff" - with an emphasis on the "building". I have to admit, I feel that my university did not properly teach me the art of creating. I only had one lecture on it and we were three students in it... I hoped I could do that a little differently with my students. – Make42 Jan 13 '18 at 20:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.