Advice to Prospective Students
I occasionally advise internship projects (at any time of year), though I rarely have funding to host interns.
I am much more likely to host you if you have funding from a scholarship, though I rarely host "self-funded" students.
If you have a strong record of publications in my area then I will make an effort to host you regardless of your funding situation.
I do not expect to advise interns during 2020 so please contact me only if your circumstances are exceptional.
Prospective PhD students:
Due to a large volume of requests I am unable to respond to individual enquiries about joining my lab. I am happy to receive (and will read) your e-mail, and will be sure to look at your application when the time comes, but it is difficult to offer useful advice before having seen the entire applicant pool. I'll respond if I have the bandwidth, but please don't expect a reply.
If you have something in particular that stands out about your file, e.g. you have publications in my area, or have worked with one of my collaborators, or have secured a competitive fellowship, then please do contact me to let me know.
Once I start reviewing applications (around early-to-mid January), I'll usually communicate with applicants via official channels, though occasionally will reach out personally if an interview is required. Again, I won't be able to respond to individual queries during this time.
Most years, I try to admit 1-2 students (which typically means making 5-6 offers). I consider all candidates who select my lab on their application (usually around 120 students, which is about typical of ML groups at UCSD). The main constraint is available funding, though this is hard to predict and depends on (among other things) student graduations, fellowship outcomes, the size of the applicant pool (etc.) not to mention my own success with grant writing. It is hard to predict these outcomes far in advance, so again I am unable to offer useful advice as to my future recruiting plans.
Selection criteria - for students applying with an MS:
The main selection criterion, given the competiveness of ML-focused labs, is relevant publications. Note that if you are applying with an MS degree, it is almost a necessity that you have published 1-2 papers in the same venues as my lab. I will mainly evaluate your application based on your publication record.
for students applying from undergrad:
Although it's important to have a strong GPA and GREs, it's rarely possible to get admitted based on grades alone. For applicants in ML it is expected that you have relevant project experience, if not published work.
for UCSD students:
Admission of UCSD students to our own PhD program (e.g. via the MS to PhD option) is possible but requires a separate application process. Note that I only admit students via this option if we have an established track record of working together, usually including published work.
Prospective MS students:
Please note that MS admissions decisions are made by an admissions committee, and as such any individual faculty member has limited influence on the outcome of your application. If you're curious about research opportunities at UCSD, see the advice below.
Students looking for project work:
I am generally happy to advise project work, though please note that it is necessary to take my class first (CSE158 or CSE258) in order to learn the necessary background material. The only exception to this is if you already know the material well from elsewhere and for some reason won't have the opportunity to take my course in time. Note also that I don't have the bandwidth to advise research projects during the same quarter I teach class (typically Fall).
Note also that it is up to you to come up with a proposal for a project you'd like to work on. Rarely do I have existing projects that I can give out to new students. I am happy to help you develop a proposal if you can give me some high level ideas of what you'd like to work on, in addition to your research background. Your proposal can be short (just a few bullet points), but should cover relevant papers, datasets, and evaluation criteria.
You might also consult my graduate students who can suggest directions for future work. If you can come up with a project with my graduate students, I am always happy to advise.
Finally, you should take project work for credit. This ensures that you'll have enough time to commit to the project. The relevant course codes for independent study / special projects are CSE199 and CSE298 (and more rarely CSE198 and CSE293). Occassionally I advise honors theses, though generally I feel that there's no reason to select this track over just doing regular project work. If your goal is to work on interesting projects, or ultimately publish something, then doing a regular "independent study" project gives you more freedom to do so without the dissertation requirement.