Switching into CS?

[Anonymous Coward]

I'm already a grad student, in an interdisciplinary program that combines machine learning, neuroscience, computational linguistics, and experimental psychology. I've realized during my time here, though, that machine learning is my true passion, independent of cognitive applications, and unfortunately I feel that the machine learning faculty in my own department are too narrow in their focus. Also, my mother's health has deteriorated to the point where I would really like to return to the Northeast and help take care of her.

Here are my stats:

2nd-year PhD student, GPA 3.93 ... the CS dept here is top 15, for what it's worth

GRE: 800M/800V/4.5W

Undergrad: Math/Economics major at Ivy, GPA 3.7

Publications: Second author on an article published in Science in the past year applying machine learning methods to neuroscientific data, second author on another paper in a very low-impact journal

Grad Coursework:

-two Foundations of AI courses in the CS department

-Information Theory course in the EE department

-Course on unsupervised learning/dimensionality reduction course in the CS department

-Course on computational neurobiology

-Course on semi-supervised learning

-Currently taking convex optimization and stochastic processes courses

Interests: Unsupervised learning, semantic modeling, neuroinformatics

What sort of schools should I aim for in the Northeast? Would it be possible for me to switch into CS or do I need more foundational CS courses?

Edited April 23, 2012 by Anonymous Coward

[jjsakurai]

I don't think you need any more CS courses. Were you admitted to any Northeastern universities that you would be fine going back to? They may be fine taking you without going through the whole admissions process. If you've a good relationship with a prof you want to work with then they might be willing to take you on as a student.

Barring those two possibilities, just apply as you did back in undegrad I guess. Definitely talk to the faculty in your department about this. You'll probably want to get recs from them anyway if you do end up applying again.

[Adamah]

I don't think you need more CS courses, but I'm not 100% on that. Machine learning isn't core CS anyway. It's really more statistics, so I don't think the professors would care whether you know your data structures if it's clear your interest is in machine learning.

I don't know if CMU counts as Northeast, but you should consider it if you're interested in ML. They have a distinct PhD for ML, and they even have a subprogram for Neural Basis of Cognition. You would also definitely want to apply to MIT, Cornell, and UMass (which has one of the best Machine Learning programs in the country).

Your situation is pretty unique. You should definitely talk to some faculty. I'm sure they'd be happy to speak with you.

[Anonymous Coward]

Thanks for your helpful replies. CMU, MIT, Cornell and UMass are definitely excellent ML programs, but don't they seem like a stretch? I guess I'm not really sure of how selective top CS programs are, but I was assuming I should aim lower than the most selective tier. If so, what is the next tier of programs I should be applying to?

It's good to know that not having taken core CS courses shouldn't hurt me too much.

Thanks guys!

[jjsakurai]

It depends a lot on your profile and your recs. Cornell and UMass, especially the latter are definitely less selective than CMU and MIT. Doesn't hurt to apply in any case. That said, if you feel CMU/MIT/Cornell/UMass will be a stretch for you, then the next tier in NE would be Brown, NYU, UPenn, UMD, JHU, and Columbia.

Edited April 28, 2012 by jjsakurai

[Adamah]

UMass is not a stretch for your profile. CMU, MIT, and Cornell might be, but you should apply anyway. Admissions can be unpredictable, both positively and negatively.

Edited May 1, 2012 by Adamah