I know questions regarding whether its possible to go from EE undergrad to CS have been answered previously, so please bear with me as I have a few distinct questions. I'm a Junior at a mid-tier state school and I've taken quite a few CS courses (such as data structures, discrete math, programming methodologies and principles, and software development).

I've been doing some reading online and I've come to find that ML/AI(esp. Computer Vision) is something that really piques my interest. I am not sure that I have the proper probability and statistics skills required to pursue a CS PhD with a focus in ML/AI. NOTE: (I do have a solid background in other math: Multi-variate Calculus, DiffEq, LinearAlg, multi-dimensional signal spaces,etc.)

  1. Do you think that it would be feasible for me to actually gain acceptance into a PhD program with this focus without any research experience in ML/AI or even CS in general? I have previous academic research experience(5 months) in EE(Semiconductors) as well as a high GPA(3.9+) and should have solid letters of recommendation. What kind of disadvantages could I face?

  2. Would you recommend doing some self-studying of ML/AI in place of continuing my current research in EE? (I'd love to do research in ML/AI but my university doesn't have research opportunities in the area for undergrads).

  3. How can I approach finding well-reputed groups/institutions in this area?

Thanks, I really appreciate any input or advice that you may have.

  • Regarding point 2, are there other universities nearby with ML groups? You could get in touch with faculty there and ask if they are willing to supervise you remotely. Jan 8, 2019 at 8:13
  • CS publishes quite a lot at conferences. Try looking at people/groups who attend major conferences like ICCV. Read the papers and see if you understand the work. Feb 9, 2019 at 1:37

3 Answers 3

  1. I think yes, it is feasible. On a purely technical side, regulations vary across countries and universities. At my university they consider eligible all applicants with previous education in "computer science and engineering". The rest undergo "additional examination", which in practice means that they check whether their actual transcripts overlap enough with our CS curriculum to be considered equivalent.

  2. Computer science is an umbrella term, so it is a really diverse area. Even if you narrow it down to "machine learning", it still doesn't help much to identify your stronger and weaker points. There is machine learning in natural language processing, in computer vision, in human behavior analysis, etc., with different requirements.

  3. Since you mentioned computer vision, I think your math background is really handy. What many such candidates lack is coding skills and general experience in dealing with programming environments. Sometimes we have to glue together a bunch of "experimental" (buggy and poorly documented) libraries and make them work. When something goes wrong, people get startled with cryptic printouts and endless error messages, as they don't even know where to begin to untangle this mess. At least, this is what I feel sometimes as I have to take personal care of the tech issues...

Sorry no comments on the "best groups", but personally I think you should approach the question the other way round. Go to Google scholar and search for papers with keywords of your interest. Read the ones you like, see who cites them and which papers they cite. Just follow the links in both directions, gather enough papers to understand (a) who is cited more often and (b) what kind of research you personally like -- after all, your personal interest in a particular type of project should matter most. And this is how you approach people: by showing genuine interest in their projects and knowledge of their publications.


In general, EEs are better at math than CS so you will probably be fine. I would not worry about prob/stat. Can pick up a course if needed. Is easier than diffyQs, etc.

More of a question will be coding ability, interest, and experience. Go code some things. (Note, I'm not telling you to take a course.)


You will be fine. EE and CS are very closely allied disciplines, so much so that at many schools, they're the same department. Here at Michigan, e.g., computer science and electrical engineering are subdepartments of a single EECS department. I teach in CS part but, like a number of my colleagues in CS, my degrees are in EE. This is very common.

You also are likely well-prepared to pursue interests in ML, vision and AI, all of which tend to require more linear algebra than statistics.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .