What are grad schools looking for?

August 7, 2008 12:41 PM

I want to get into an excellent CS grad school, but I have no idea what I'm supposed to do, or what they're looking for.

I'm currently a computer science major at UC Berkeley (starting junior year), but I don't know what I need to do to get into grad school. I haven't done much in the past 2 years, unforunately - no jobs, no internships, no interesting projets - and I only have a B average in my technical courses. In fact, I feel my only defining characteristic is that I'm double majoring in music.

Aside from improving my grades (which I'll try to do next semester), what should I do to become a more viable candidate for the top CS grad schools (particularly in England), and where can I find more information on this topic? What else will I have to do to get into grad school? (i.e., letters of rec, etc.)

(I feel one of my main problems is that I only started programming a year ago, so I don't really know how to do much yet.)

posted by anonymous to education (12 comments total)

CS grad schools are interested in research and specialization. Get specialized by working on specific projects with a professor or researcher. Perhaps you can find projects where your music background will come in handy (computational music analysis?), but find a way to help out.
posted by demiurge at 1:01 PM on August 7, 2008

Peruse your uni's cs/music faculty listing. Find a few doing something you find interesting. Email them and ask if and how you can get involved. Profit.
posted by norabarnacl3 at 1:05 PM on August 7, 2008

I think it's a little too early for you to know whether grad school is really what you want. What you should do first is to try to make your next two years in college as much like grad school as possible.  This means:

* Start actively seeking out faculty advisors and developing relationships with them, both from your courses and in research areas that interest you
* Become a research assistant on one or several projects; get as much research work under your belt as you can
* Start working on side projects of your own if possible; at a minimum start actively reading recent published research, and thinking about what kinds of research work interest you
* Do as much programming as you can in as many languages as you can

In general, doing research at whatever level you can, and seeking out and becoming informed about the research fields that interest you, will both make you a better candidate for grad school and help you decide whether it's right for you. Be proactive; talk to as many grad students and faculty as you can, for as long as possible, about their work and their experiences. There are great programs out there in computer music, but you don't have to do that just because you can.

Phil Agre, a computer scientist, has some great general advice for prospective grad students.
posted by RogerB at 1:10 PM on August 7, 2008

Why do you want to get into grad school? What are you interested, what would you like to do or study?

Once you answer these questions, just start doing it. Read up on your own; talk to professors that do related work, play around with personal projects, etc. CS grad schools (I'm talking PhD programs) tend to use grades as filters -- you need decent grades, but it's not at all enough to get into a top program. Rather, what they want to see is that you like the field and have been involved in projects/research.

Also, don't get involved in projects/research just to get into grad school -- rather do them in order to start achieving what you want out of grad school, and things will come naturally. You'll get a better idea of whether you really want a PhD (this applies less to a MS), you'll learn about some of the big names and their work, and you'll make yourself a better candidate in the process. You may find that your interests are completely different than you currently think they are.
posted by bsdfish at 1:13 PM on August 7, 2008

Yup. Take graduate classes and talk to as many faculty members as you can - your best bet is getting on board with one of them as a research assistant, since their connections and guidance is the most valuable thing you have access to at this point in your career.

Grad school isn't what you do if you don't know what else to do, though. It's a royal pain and not really all that useful unless you need a faculty position somewhere. You have plenty of time to decide and fortunately, the networking route I and others have mentioned is the best path for both graduate school and working straight out of undergrad.
posted by kcm at 1:22 PM on August 7, 2008

I went to a good CS grad school (in the US) with pretty much no CS background whatsoever. It's probably harder to get in now, but I'd nth the research suggestions. If you're at a research university, there should be something for undergrads to work on.

And as a CS lecturer, I would tell you to go to office hours. The best recommendation letters I've written have been for students who do well in my class, but are often not in the top 10 percent, and who come to office hours all the time. Then I can write something about the effort they put in, about some insightful comment they made that shows that this one bad grade they got on a project doesn't really mean anything. (Even if 90 percent of the comments a student makes are dumb, I'll forget all but the intelligent ones.) And if you do well in a class and have a good rapport with the instructor, consider asking for the recommendation right after, or before, the class is over, so it's all fresh in the instructor's mind. Even if you don't need it right then, s/he will probably have it on file for a while.
posted by transona5 at 1:27 PM on August 7, 2008

If you are only interested in a Master's and not a PhD then Berkeley has a 5-year combined Bachelors/Masters program in CS. But that's not in England.
posted by GuyZero at 1:40 PM on August 7, 2008

As someone who got their Masters in CS about a year ago, I think it's weird that you don't mention why you want to go to graduate school. Maybe you just didn't think to mention it, but if you can't communicate clearly why you're interested in grad school, you've got to figure that out before anything else.

I opted for grad school after spending over two years working in a graduate research lab on an NSF REU grant (which is a great way to do do undergrad research and not starve/drown in debt, btw). I enjoyed it to the point where I wanted to keep doing it after I graduated.

There's great advice upthread (esp. finding research positions and getting to know professors), but first and foremost, you've got to know why you want to go to grad school. If you're not sure, talk to some current grad students about it. Grad school has perks, but it has lots of disadvantages too.
posted by Nelsormensch at 3:03 PM on August 7, 2008

+1 for getting involved with a research program. More specifically, figure out what specific subarea of CS you're interested in (AI, theory, databases, networks, etc) and try to get involved with a research program in that area.

(I did my PhD in CS, and am now a professor, so I've seen both sides of this.)

Grad school is all about doing research, so the application process is focused on looking for people who can come up with smart ideas, work independently, and carry things through to completion. Having done that sort of thing as an undergrad (and having a letter from a professor who's supervised your work) is a big plus. Getting good grades is (almost always) a necessary, but not sufficient, condition to get into grad school. What will make you stand out are publications, if possible, and strong letters from professors who can talk about your research and independent thinking skills. Side projects are also good. Do something on your own, or get involved with an open-source project.

A few things to add:
- Grad school is also a long haul, and you really have to enjoy what you're doing to make it through. If you do, it's great, but if you don't, you'll be miserable. (you might be miserable anyway...) So think carefully about the area you're interested in - do you want to spend the bulk of your cycles for the next few years thinking about it? I also agree that you should think about what you want to do with a PhD - it will open some doors (academia and research) and close others (entry-level software engineer, for example).

- The good news about being at Berkeley is that there's a lot of research going on, so it shouldn't be too hard to find an area or project you're interested in. The bad news is that there are lots and lots of people who want to do research, and so getting involved with a project can be difficult. Some professors are not that excited about working with undergraduates, because they take more training and supervision, and don't produce as much bang for the buck. So I'd suggest that you be tenacious and prove yourself - if there's a project or professor you're interested in, ask if you can sit in on their meetings or go to their reading groups. Be willing to do the yucky grunt work at first. Also, follow through - if you say you're going to do something, be sure you do it. It seems simple, but this is a huge problem for a lot of students. Also, talk with grad students (such as your TAs) in your area of interest; they can give you an idea of what schools are excellent in that particular area, what the job prospects are like, how different professors are to work for, and one version of what the grad school experience is like.

- The PhD process is a little different in Europe than it is in the US. The programs are typically shorter (3 years rather than 5, for example) and your thesis might be less in-depth. By itself, this is fine, but it might mean that you'd need to do a postdoc if you wanted to come back to the US for an academic job. You'll also want to look at specific schools in England to find out what the residency requirements are (do you need to be a UK or EU resident to get financial support?).
posted by chbrooks at 4:21 PM on August 7, 2008

Can you think of a way you'd like to interweave your interests in CS and Music?

Assuming you have interests in CS of course.

DirectMusic may have been before your time, but it was an interesting direction in computational music creation.
posted by yort at 6:08 PM on August 7, 2008

Coming from the other side of CS and music (i.e. the music side) I'll say that there are quite a few interdisciplinary people and departments around as long as you want to get involved in the composition and/or improvisation side of music, indeed I know quite a few people who are involved in this. Feel free to MeFiMail me if you want.
posted by ob at 8:40 PM on August 7, 2008

You do know about CNMAT, right? They're small enough that the teachers over there have tons of time for undergrads. I took a MAX/MSP course offered by CNMAT while I was at Berkeley and it was great. They'd probably love to have someone like you help out in the labs and would probably let you play with some of their awesome equipment if you're nice (eight-channel sound system, anyone?).
posted by fishfucker at 9:08 PM on August 7, 2008