3

I'm in a situation where I've probably bitten off more than I can chew. I decided to do my master's thesis on an original (as in, invented by me) machine learning algorithm, which I believe (and my advisor agrees) is interesting and might have something to contribute to the academia.

As a part of the thesis I wanted to implement the algorithm. However, implementing it is turning out to be a major pain. I've been working on the implementation for several months and the end is still nowhere in sight. Due to this, I'd like to change my thesis to be simply a description about the algorithm and comparison of it to existing, similar algorithms.

So my question is, is it feasible to have an original, unproven and untested idea as the main focus of your thesis? This feels iffy to me: master's thesis is supposed to be a sign of academic maturity, but nothing seems more unacademic to me than publishing an untested and unproven idea in the form of thesis, and not even having a working implementation of it.

I'd really appreciate advise, dragging this thesis is really getting the better of me. Thanks.

closed as off-topic by Davidmh, scaaahu, Bob Brown, Brian Tompsett - 汤莱恩, gman Apr 25 '16 at 20:00

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "The answer to this question strongly depends on individual factors such as a certain person’s preferences, a given institution’s regulations, the exact contents of your work or your personal values. Thus only someone familiar can answer this question and it cannot be generalised to apply to others. (See this discussion for more info.)" – Davidmh, scaaahu, Bob Brown, Brian Tompsett - 汤莱恩, gman
If this question can be reworded to fit the rules in the help center, please edit the question.

  • 3
    I'd like to change my thesis to be simply a description about the algorithm and comparison of it to existing, similar algorithms. — How will you compare your algorithm to existing ones? – Mad Jack Apr 24 '16 at 23:57
  • 1
    What does your supervisor say? He should know how to redirect your project, and what is expected of you. – Davidmh Apr 25 '16 at 11:08
  • Could you instead take one aspect of you new original idea and modify an existing (implemented) algorithm to apply that aspect? Or could you get help implementing your algorithm so you could get it done in time? I assume you're being judged on the idea and comparison not the implementation. – Rup Apr 25 '16 at 11:36
2

Write a chapter of your thesis in the shape of a review paper, which is good enough to submit for publication. Submit it. Write another chapter which builds on the review and uses that as an argument for the derivation of a new algorithm. Present the algorithm. In the discussion, highlight the steps needed to implement the algorithm and point out the problems encountered when you tried implementing it yourself. That's a MSc for me.

1

If you have made an original contribution to your field and added to what the community knows, that would seem to me to be sufficient for a master's thesis. Many theses at this level do not even reach that level, and students essentially end up writing review papers for their theses. On the other hand, what you have done does not necessarily reach the standards for a particularly good masters-level thesis. If you are planning to continue in a doctoral program, what will matter will not be whether your thesis is adequate, but whether it is strong enough that your thesis supervisor will give you a very strong letter of recommendation. If your advisor feels that you could have implemented the algorithm, but instead you gave up, that will affect the strength of the recommendation they may give you.

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