Matching students to advisors is done relatively early at Edinburgh, and students usually know who they are going to work with before they join the school as Ph.D. students. If you are interested in working with me, please explicitly mention that in your application, so that it is also routed to me. Also it would be good if you separately get in touch with me (see question below).

Send me an email (scohen [strudel] inf.ed.ac.uk). Please provide some background about yourself, your current academic status and the reason you think my research is a good fit for your interests. A CV with a list of publications (if there are any), or alternatively, a pointer to your home page, would also be helpful. If you are visiting Edinburgh or you are already a student here, setting up a meeting with me is the best way to see if we are a good match for working together. If you want to get the big picture of what I do, check out my publications page here. The more recent a publication is, the more representative it is of my current interests (roughly -- there is older work I would like to revisit).

I am looking for students who have an interest in natural language processing or machine learning. You should also have relatively strong programming skills and mathematical maturity. Right now I am especially interested in applying linear algebra to statistical learning of language through spectral learning. In general, I am interested in structured prediction for NLP and machine learning. I am also interested in the use of neural networks for structured prediction and generation.

ELLIS offers a federated search mechanism that connects universities to applicants and potential supervisors to other potential supervisors. At the moment, it does not provide funding, and its recruitment cycle is a bit different than the one in some universities. I highly recommend that you apply also to Edinburgh separately even if you have already applied for a PhD through ELLIS -- this is something you would have to do anyway, even if recruited through ELLIS. Let me know by email that you also applied through ELLIS.

The funding sources for the CDT and the ILCC programme differ. This means that you could potentially get accepted to one but not the other. If you can afford the time and cost of applying to both programmes, I highly recommend you do that. Make sure you are aware of the differences between the two programmes.

You are more than welcome to send me an email or meet with me to see if I have a project which suits your skills and interests. For the Master's and undergraduate programme, the projects I currently propose for students are listed on the Informatics webpages. However, on certain occasions, when I am convinced the student can design a project of his/her own, I am happy to listen to student suggestions for projects.

I do that every so often, but quite rarely. If you want to work an MRes with me, contact me, and explain why you think it is a better choice for you than a PhD. I do expect MRes students to be in a similar level, at the very least, as beginning PhD students.
Note from December 2022: I am looking for one MRes student for a project to develop non-traditional training algorithms for neural networks. You would need to have mathematical maturity, and experience in just coding ML or Deep Learning models is not sufficient. The code you would develop is likely to be relatively simple. If interested, contact me, as the time of admission is flexible for the MRes degree.

There are quite a few advice pages on why, when, where and how to get a PhD. Maybe one day I will write about my own perspective, but for now here are some useful resources:

• Karl Stratos' advice page.
• Kevin Gimpel's advice page.
• Mor Harchol-Balter's advice page.
• Noah Smith's advice page.

As everything else in life, you have to take it all with a grain of salt, and see what fits your situation and views.