How to approach thesis writing in MSCS

By Michael Soltys
Updated September 12, 2018

Writing a masters thesis can seem a daunting task. This short document offers some guidance to make the process more manageable. Please keep in mind that you should follow the lead of your adviser who may have different suggestions; this write up is to serve as informal guidance.

Most theses take a form similar to the following:

  1. Introduction
  2. Background
  3. Contribution
  4. Experiments and justification
  5. Conclusion and future work

The first chapter, Introduction, contains a short motivation to the topic of the thesis, and a list of 2 or 3 (or more) contributions of the author (you). These contributions are specific; that is, “introducing machine learning to self-driving cars” is too broad, but “implementing algorithm X in C++ and showing that it is twice as fast as the best contender on inputs over the binary alphabet” is a specific contribution.

The second chapter, Background, presents all the material needed to the reading of the thesis. That is, all the definitions, results that the thesis builds on and the terminology are presented here. The aim is to ensure that the reader can understand your contributions precisely. This chapter also contains a literature review (which sometimes is done instead in the Introduction). This is very important as it places your thesis in the context of other work done in the field, and justifies the importance of the contributions.

The third chapter, Contributions, is the heart of your thesis. It consists in a detailed description of the contributions that you first listed in the Introduction. This chapter is the technical work of the thesis, and if the contribution is conceptual it will include justifications such as mathematical proofs, and if it is a program it will contain snippets of code and links to GitHub repositories. This is the chapter that will be read in detail by someone who may want to make use of your contributions.

The fourth chapter, Experiments and justification, is contingent on what type of contributions you described in the third chapter. If your contributions were purely conceptual (e.g., a new proof of correctness of a known algorithm – say, a stronger proof that establishes that the algorithm works on more cases than previously asserted), then you may not need to run an experiment. But most theses in our field contain a combination of conceptual contributions and programs; the value of code is justified by running experiments on some fixed platform where your program is compared to the main contenders in the field.

Finally, the fifth chapter, Conclusion and future work, repeats the value of your contributions, and presents the reader with possible future developments. Rarely a thesis is a final word on a subject, and a robust future work section shows the depth of your understanding of the project, and offers enticing prospects for other masters students who might be interested in the same area. It also ensures your legacy, in that the future work will be done by researchers who will be citing your work as the source of their inspiration.

With the above in mind, as you are sitting down to think about your thesis work, start by selecting a broad field of interest, and as you read articles and books in the area, narrow down as much as possible a possible topic of research. As you start thinking and working on the topic select two or three very specific contributions that you wish to make. Your adviser will be helping you in this process.

For your reference, the timeline below is a countdown to your graduation date (either May or December). Keep in mind that you are to follow the lead of your advisor who may have a different timeline in mind, more appropriate to your circumstances.

  1. Two and a half months: submit your thesis to your adviser who willtake about two weeks to read it and decide if it can be submitted to the committee. If the adviser decides that the thesis is not ready for a defense, the process has to be restarted following consultation with the adviser.
  2. Two months: submit your thesis to the committee.
  3. One month: defend your thesis.
  4. Two weeks: re-submit your thesis. This is assuming that you passed, and the committee requested minor modifications. In general, minor modifications have to do with presentation issues and small improvements (e.g., run twice as many experiments), whereas a major modification indicates that there are substantial issues with your thesis (wrong results, incorrect implementation of major components, etc.). In the case of major modifications the committee needs to approve the thesis, whereas in the case of minor modifications only the advisor needs to approve the final version.
  5. Submit thesis with signatures page to the director of the MSCS graduate program (Michael Soltys currently) who in turn submits it to the school of graduate studies (EU).

Defending your thesis

Your defense will consist of a seminar presentation, where you should talk for about 45 minutes, with 15 minutes for questions from the audience, and up to half an hour of questions from your committee. Questions may be asked during your presentation as well. Keep in mind that, depending on the level of technicality, each slides takes 2 to 3 minutes to explain, thus you should aim at about 30 slides. Your presentation should be polished, it should cover your contributions, some background, and overview of the importance of your chosen problem, and one discussion that is more technical to show the depth of your understanding.

After the committee is satisfied with your answers, you will be asked to leave the room (the audience as well, only the committee will remain). The committee will then deliberate, and establish one of the following:

  1. Fail
  2. Pass with major revisions (there is a substantial issue with your thesis; your thesis needs to be resubmitted to the entire committee for review)
  3. Pass with minor revisions (stylistic issues, improved or added graphs, etc., but the results are sound and you only need to resubmit your thesis to you advisor)
  4. Pass without revisions