Chances of getting into a CS PhD program from a math/engineering background?

Question

I have an undergraduate degree (B.E.) in Biomedical Engineering. Right now, I am working on a M.S. in Applied Mathematics and Statistics and will be done pretty soon. I have very minimal research experience, and no papers published. I will be working in industry for the next year (since I didn't apply for any PhD programs this application season). However, I would like to get into a PhD program as soon as I can.

My interests are in the field of AI and Machine Learning. From the research I have done, most AI/ML research is done by Computer Science departments; so that is where I must head. What are my chances of getting into a good AI/ML program coming from a different background and no relevant research experience (although I have no direct research experience, I follow the literature and learn as much as I can on AI/ML in my free time)? What can I do to best increase my shots of getting into a top CS department known for AI (Stanford, Cornell, etc)?

Instead of working for a year, maybe I should seek a research assistantship at an AI lab (less pay, but if it can get me into a top program then so be it)?

Any advice is appreciated!

Answer 1

1. High GRE scores open the door (quantitative reasoning and to a lesser extent analytical writing).
2. Math / Statistics background will raise eyebrows.
3. Creating a portfolio of projects to show the department that you can in fact code, will get you assistantships.
4. Be pro-active but not a pest.

When applying to a university, you might want to peruse the various web pages of faculty to see what fields they are publishing in. If you find a field that piques your interest, or that you've done some similar work in, contact the professor directly and seek advice. Be careful though. I get emails all the time with prospective students phishing for assistantships. I can tell when it's a bulk / canned email that was sent out to a ton of faculty just looking for someone to bite.

I would start with a personal tone:

Hello Dr. ______________ I was looking at your work on _____________ and wanted to let you know that I am very interested in this research topic.

It probably would be even better if you could tell them that you already applied to their university, but I understand that you would go broke with all the application fee's if you applied to every university that peeked your interest.

Briefly explain what your seeking (cool research, with possible assistantship). Be patient, you might not get a fast response, if ever. These guys are busy, and probably getting a bunch of emails as well. Start early. At least a full 12 - 18 months out. This way you can narrow down your choices and not have to worry about application deadlines coming up too quickly.

Communication skills can be big. If your comfortable in front of a classroom, and your english skills are good, maybe looking for a teaching assistant position could increase your chances. This does mean you won't as much research, but at least your getting your foot in the door. This will give you more opportunity to show a particular faculty member what your worth.

Answer 2

Terry has a great answer, so I'll just add a bit more...

The biggest hurdles are your math, logic, and coding skills. If you can prove you have these, then you'll be in good shape. CS, on a theoretical level, is far less coding than you might think. I had professors in CS who didn't even have computers in their office, and made a point to never use/require computers in the classroom. Simply put, there's a lot of thinking, work, and research to be done on paper before you even attempt to program.

Of course, since you're into AI, that means you'll need to code quite a bit. I don't know how much you know about AI, but I would recommend Artificial Intelligence: A Modern Approach which, in my experience, is the de-facto entry to AI in academia. It's usually split between two courses, sometimes one of which is undergrad, and the other grad level. It has used various languages over the years, but the two most notable ones are Lisp, which deals with list comprehensions, and Java, which is a typical Object-Oriented language. Having experience in both would be very beneficial for you on your CV as well as when you actually start working/researching.

Answer 3

Honestly if you can prove yourself in the field then you should have no problem. I would suggest reaching out to various universities and trying to prove your knowledge. For instance my math advisor has a bachelors in engineering, a masters in economics, and a phd in math. It just depends on how much you know and how much you want it. It is most definitely possible.