Applicant Q&As

Read some common enquiries and our responses regarding the application process and the nature of our programme.

Application Stage

Research Proposal

We want to see evidence that you have stopped to think about what it involves to apply for a PhD and what it would mean, and that you have some understanding as to the area that you’re looking to apply into, and what sort of research would interest you.

A research programme isn't just applying for a degree or applying for something where you have a generic interest. It's about having an interest in being able to actually push the frontier forward in a particular field. The research proposal enables us to see that you have thought and really care about that. Doing research on something you don't care about isn't going to work. 

Your research proposal should outline the background research area that you're considering. What is it that is going to be important in there? Maybe you have identified one or two problems in that area that you think are good problems to be addressed? You might not perfectly know how to address those problems yet, but you can still work out if there are problems that you care about. Maybe you know of some background research papers or things that you've looked at that relate to that particular area. If you have some suggestions as to particular ways or particular tools that you might bring to tackle those problems, that's good too. 

It is about what you’re looking towards doing over that course of the PhD. Now our candidates obviously are coming from very different backgrounds. We’re getting applications from mathematicians, physicists, computer scientists, people who did degrees a while ago and had been working in companies for a while, and they’re in very different positions. So it is very much a matter of you starting from where you are at and trying to build on that.

More guidance on the research proposal


We would like to see some level of detail; vague statements are often not helpful. The recommended length is between 2 and 5 pages, references and images excluded.

Note that you are not committing yourself to working on that project during the course of the PhD. It is just used as an exemplar that helps you to express the things that you care about, that allows us to engage that, match that up, and understand where you’re coming from.

More guidance on the research proposal


It’s possible to engage the prospective supervisors to help or get feedback on your proposal, but we want the proposal to come from you, not from them. It is about what you want to do now as you write that proposal. 


References

We do not place a huge emphasis on references but we do look at them and take into account what referees say. They are helpful to get some understanding of your previous research experience and the skills you are coming to us with and which can be leveraged over the course of your PhD.


There is a preference for an academic referee only in the sense that it is more typical for academic referees to go into a little more detail about your academic and research capabilities. A bland industry reference saying they know you working in the company for this length of time is generally less helpful. On the other hand, if the referee is someone who knows you well, who you’ve worked with in a company setting and who is willing to make comment about your particular skills, that can also be helpful.

The key question when choosing your referees is: will they be able to write a recommendation letter that gets across your skills (in the context of academia or a company)?


Candidates' Profiles

No, it is not a problem. We get quite a few applications from people who are moving back into or coming into doing research after having been working in a company. That perspective actually provides a lot of added value. In fact, people who have been off working and made the decision to then go into doing a PhD have often given that serious thought and are strongly committed. However, it does involve a little bit of a switch and to put in some background work to try and get up to scratch as to what you might care about within that PhD programme. Talk to people, and work out what things interest you. If you have done this background work, working elsewhere and the skills you learnt from that experience are very valuable things to bring into your PhD study.


Supervisors

Any primary supervisor in the University of Edinburgh could work as long as:

  1. there is an active co-supervisor that is based in the School of Informatics and 
  2. that the research topic is within the CDT’s remit. 

Supervisors listed on our website are only a few examples from the School of Informatics (supervisors who would particularly like to attract applicants) but there is a wide range of supervisors you can approach within the University.


See our guidance to contact a prospective supervisor: Application Process and Guidance | School of Informatics

Academics are very busy people and some receive hundreds of emails every month from prospective PhD students hence it can take a while until they are able to get back in touch with you. Be patient and keep prompting them every two weeks or so.

If after several weeks you have not had a reply, you can email the CDT admin team at mlsystems-enquiries@inf.ed.ac.uk letting us know who you have contacted and we can chase up the supervisor(s) for you.

If the deadline for application is getting closer, you can still submit your application and nominate the supervisor. We will forward your application onto them for assessment.


Programme

The CDT’s scope is to focus on Machine Learning and AI methods that are effectively working when deployed, essentially by bridging the gap between ML and Systems and bringing together people from these two disciplines. We see the whole ML and Systems as a spectrum. The CDT aims at recruiting a good balance of people who are coming from the ML side of things, solving ML problems, but doing that with an awareness of what's required to actually get those working in a real-world setting, and people coming from the Computer Systems side, solving Systems problems, but with a knowledge of ML methods to identify how they can work best when deployed. Within the CDT, there will be students who will work more on the ML side and others who will work more on the Systems aspect but we expect research projects to include both elements and interactions between students and expertise.

Projects with a main ML component are based on fundamental ML and AI questions. For example, that could be addressing continual learning error to adapt the ML methods to the changes that they see in the real world (e.g. devices or material evolution) or working on how we combine learning from different sources, how we transfer learning from one setting to another.

There are also projects looking at particular AI applications, focusing on solving an application, not just by developing the ML method but also by getting it working. For example, if the application is a real time application, it’s not enough to build a wonderful method that isn’t able to run in real time. You have to think about what the real constraints are that the world provides.


Part-time studentships are only available to students not needing a visa to study in the UK (mostly students eligible for Home fees or with British citizenship). This is because the University of Edinburgh does not sponsor Student visas for part-time studies. 

There are also some important aspects to know about studying a part-time PhD:

  • A part-time PhD is strictly 0.5 FTE, i.e. 17.5 hours per week, no more no less (there is no flexibility on this)
  • The monthly stipend is therefore halved and the duration of the PhD is extended to 8 years (vs 4 years for full-time study) 
  • Students in full-time employment are not eligible for an award of any kind from the UKRI
  • Students in part time employment are eligible for a part time award. The University does not recommend studying/working more than 44 hours a week (all inclusive) with a recommended average between 35 and 40 hours. This is to ensure that enough time and of quality is spent on your studies and that your work-life balance and wellbeing is not affected (this is especially important for an 8-year programme). 
  • The suitability of a research project being done part-time must be checked with the chosen supervisor from earlier on since studying part-time might not be suited to some fast-moving areas or topics, given the longer study timing.
  • The CDT is a cohort-based programme with some credit-bearing courses meaning that while there are adjustments possible to fit part-time studies, there are a number of courses, training and events that students have to attend and that cannot be rescheduled. This therefore requires some flexibility of the other commitments (be it employment, caring duties or other commitments).

The process to apply for part-time studies is similar to full-time studies, you just have to select the part-time programme from the degree finder to trigger the correct application.


Your research proposal does not necessarily constitute your PhD project (partly or wholly). Your PhD project will be defined with your supervisor during your first year.


The programme is strictly 4 years (FTE) which is also the duration of the funding. You have the option to go into a fifth year (writing-up year) to finish and submit your final thesis however, the fifth year is not funded. 

The earliest you can submit your final thesis is 3 months before the end of the fourth year. It is quite rare that students finish before the end of the 4th year. There is a lot of opportunities in the latest period to really build your profile (e.g. conferences, publications etc.)


The MLSystems CDT is a four-year PhD degree. This is not a PhD with integrated Master degree (also called “1+3”). However, MLSystems is a “PhD with Integrated Studies” meaning that there is a number of credits (total 720) to be obtained during the course of the PhD. This includes some credit-bearing courses (mostly concentrated in the first year) and some credit-bearing projects, in addition to the final thesis.


There are many opportunities of industry engagement within the MLSystems CDT.

  • Student working with partners within their research project (or beside their research on ad-hoc industry-led projects);
  • Industry-led Mini-projects that all students work on in groups (Year 1);
  • Entrepreneurship and Company Day (annual CDT event) with talks and roundtables to interact with the CDT’s industry partners;
  • Industry-funded studentships (these will be advertised separately from the main recruitment cycle);
  • Internships (Year 3), 4-6 months, distinct from the PhD research.

MLSystems students are encouraged to publish over the course of their PhD study, and generally they will do so in good venues that help progress that research. This helps build the students’ profile and # CV. It also provides interactions with others and helps them to establish themselves as leading researchers in a particular area. Supervisors spend a lot of their time working with students, helping them in their ability to write these sorts of publications. 

That said, publications, or acceptance of publications, is a fairly random process. Students can find that they do good work but still get their papers rejected to not necessarily good reasons sometimes. Therefore, at the same time as encouraging students to go for publications, the CDT does not evaluate students on the basis of publications in any way. It is all about doing good work and hopefully good publications arrive as a result because you put in the effort to write things well to describe things well. Research that is not communicated is not research in the sense that you need to be able to make your research known to others as part of that process. Publications are one way of doing that, but not the only way.