How to get First Class Honours

Honours is the 4th year of the Bachelors degree in Australia, South Africa and New-Zealand. It is usually an optional year as you can choose to graduate after three years. This guide will help the reader decide if they want to complete an Honours year. It will also provide advice for achieving First Class Honours. Although examples come from the Computer Science (CS) major, this guide may be useful for Honours in a different subject.

Honours in CS at the university I attended is structured as 50% coursework with 2 courses a semester and 50% research with a full year research project. The Honours structure depends on the major, some majors such as Biology only have a research component. I found the workload to be high, but I put a lot of time and effort into the project.

Is Honours worth it?

I enjoyed the research component of the Honours year, and thus Honours overall. For the research experience alone I think Honours is worth studying even if you plan to enter industry after graduating. Honours allows you to understand how new knowledge is created or theoretical problems are solved. The research project makes the program unique and is something you are not exposed to in industry.

Honours opens up the possibility of further study such as a PhD. You will also be able to meet other ambitious students and network. My observation was that the cohort of about 40 honours students contained some of the best students in CS. Completing the dissertation should improve your scientific writing skills. Good writing is valuable both in industry and research, where you need to communicate your ideas clearly.

The Research Component

You will first need to find a supervisor and research topic. You can meet with several researchers face to face to get a sense of whether you would get along with them. Communication styles and research guidance will differ between supervisors. Some supervisors are willing to give more guidance whereas others are more hands off and expect you to think of everything yourself.

Usually researchers have a list of projects they have proposed that you can select from. If you have another research topic in mind you can propose your own project. My supervisor only had the research topic set but no specific problem. Completing the 6000 word literature review due at the end of the first semester will help you identify a problem. The literature review ensures you are familiar with the research area and what has been studied in your topic of interest. It is useful to describe limitations of past studies in your literature review. From the literature review, in particular the limitations of past papers, I identified potential research problems.

Reading other papers becomes easier over the year as you get more familiar with the research area and improve your knowledge. It is more useful at the start of honours to be selective about what you read in a paper, rather than reading a paper end to end. This means understanding the core novelty in the article, usually in the methodology section. It is advisable to be critical of papers too, a paper may be flawed or not reproducible.

Once you have chosen a specific problem you can create investigation aims. I used investigation aims but it may be the case a hypothesis is more suited to your problem. There may be multiple aims you would like to investigate. The scope of the project should be discussed with your supervisor, but it is likely to be small so that it can be completed within the time frame. At this stage it may be helpful to visualise the contributions of a successful project that achieved your investigation aims. There are a few dimensions the contributions of the project can be visualised.

The first is the novelty of the project. How novel are the investigation aims/methodology? How similar is the research to past publications? How much new insight does the project provide? The second is the difficulty of the investigation aims or the research problem. How hard was it to address the aims or solve the problem and why? The third is how usable is the result? Can others use anything produced in the project such as models, code or datasets?

My dissertation had a 15,000 word limit and was due in the last week of the second semester. Progress on the research problem was very non-linear with numerous ups and downs. I had to change my investigation aims after reaching hurdles in my experimentation. At some point you may no longer believe your approach is a good path to go down. You can just change the direction of the project, provided you have enough time. Failed experiments can be reported in the dissertation as they demonstrate your research effort. A dissertation with only failed experiments that is well written can score very high, though it may not be publishable.

One way to speed up your experimentation is to make use of more computational resources. I was running 10 GPU’s simultaneously at one point as well as overnight just to get my results quicker. Let the computer do the work then focus on writing. The winter break is another good opportunity to make progress on the project. LLM’s can be also be used to speed up your progress. From formatting your LaTeX document, writing code for you, helping you come up with experimentation aims, providing feedback on your writing and helping you understand past papers.

If you are using Overleaf you may find it useful to use git and your favourite text editor on your computer rather than the website. You can then run terminal coding agents on your work with the benefit of git to undo changes. A second git remote to your own Github repo can be added so you have your own copy of your work. I used a second remote for pushing only so that my Github repo also contained the changes.

The Coursework

The coursework was valuable as I gained computer vision and deep learning skills and knowledge. The only downside was the machine learning and deep learning courses used a practical and nonrigorous book in Hands on Machine Learning.

Achieving First Class

Achieving a good grade in coursework is no different to earlier years. It is important to aim for high coursework marks, as you cannot be certain how the research project will be received. When writing the dissertation you should clearly state every single thing you did yourself. You should highlight your work rather than the research you build on top of. It may be helpful to think “you are a capable computer scientist” when writing, so that you focus on demonstrating this in the dissertation. The dissertation is the only artifact that is graded, the code and datasets are not (though they may be checked). So your effort should go into making sure the dissertation is well written and highlights your work. Comprehensive experimentation can also make your dissertation stand out. The dissertation needs to impress the marker in order to score high.

Before starting the program I had read many dissertations so I was able to get an idea of their contributions and how they scored. The dissertations that scored high were well written. The dissertations that were clearly second class had many poorly written sections, very little written in the discussion section and poor formatting. I re-wrote many paragraphs, some multiple times, to make sure they were as clear as they could be for the reader to understand. This follows from the idea that writing that sounds good is more likely to be correct.

I was able to spend more time re-writing paragraphs as I wrote a draft of each section as soon as I was able to. Even if you are not able to write a section yet eg the discussion, it may be helpful to write down some bullet points for what you think you should discuss. After receiving feedback on my submission I had multiple reviewers comment that a part of my methodology was not clear. I had made a conscious decision to only include my work in the methods section and exclude techniques I used that were not mine. So it is important to aim for clarity and include everything relevant in each section but also stating what is your work and what is not.

Motivation

I am going to provide some resources that motivated me and may also motivate you. The first is the book Psycho-cybernetics by Maxwell Maltz which I read just before starting the course. The idea is to have a clear goal in your mind so that your subconscious can begin planning to achieve it. I set the goal of achieving First Class Honours, which I obtained. If you set First Class as a goal, you can then think of the goal each day so you are motivated to work hard and achieve it.

When I was completing the research, everyday I would have a strong desire to spend time working on the dissertation. Even after I had submitted the dissertation, inside of me I still had a strong desire to open my neovim text editor and keep typing. The desire persisted for a while after the submission deadline passed. That’s how motivated I was to achieve my goal. The honours year is worth it, to be able to come away with an understanding of the research process. If you give your absolute best effort then you will have nothing to worry about.