AI: MIT or Berkeley?

[NineIsPrime]

I agree that the reputation and quality of the two schools are comparable. I think the most appropriate thing to do is to visit both schools and see at which one you feel more at home. Talk to the graduate students from both places and see where they seem happier. I go to Berkeley for undergrad and I LOVE the city. Of all the college towns I've been to, Berkeley is the best by a significant margin.

Here's what all know about profs at Berkeley:

Michael I. Jordan is a living legend in AI/ML; your future in academia is fully secure if you can become his student; but he seems already super-busy with the legion of students he has so far.

Stuart Russell is probably just as renowned in AI/ML; but he is currently the department chair and thus quite busy.

There's also Peter Bartlett and Martin Wainwright for ML, Jitendra Malik for vision, Dan Klein for NLP, Pieter Abbeel for Robotics, Richard Karp for Bioinformatics (he no longer does complexity theory I hear), etc. These guys are all rock-solid except for Richard Karp, no one has heard of him

Beware though that the Berkeley AI Prelim is extremely hard. I think about 30% of the people pass the AI prelim and you only get 2 tries (3 if your advisor likes you a lot). So quite a bit of AI grad students never end up passing the prelim and have to change to another field or walk away with MS.

On a side note, is it general consensus that CMU is better for Machine Learning than Berkeley?

[bernard]

These guys are all rock-solid except for Richard Karp, no one has heard of him

You gotta be joking. He's got a Turing under his belt.

[fanta]

I agree that the reputation and quality of the two schools are comparable. I think the most appropriate thing to do is to visit both schools and see at which one you feel more at home. Talk to the graduate students from both places and see where they seem happier. I go to Berkeley for undergrad and I LOVE the city. Of all the college towns I've been to, Berkeley is the best by a significant margin.

Here's what all know about profs at Berkeley:

Michael I. Jordan is a living legend in AI/ML; your future in academia is fully secure if you can become his student; but he seems already super-busy with the legion of students he has so far.

Stuart Russell is probably just as renowned in AI/ML; but he is currently the department chair and thus quite busy.

There's also Peter Bartlett and Martin Wainwright for ML, Jitendra Malik for vision, Dan Klein for NLP, Pieter Abbeel for Robotics, Richard Karp for Bioinformatics (he no longer does complexity theory I hear), etc. These guys are all rock-solid except for Richard Karp, no one has heard of him

Beware though that the Berkeley AI Prelim is extremely hard. I think about 30% of the people pass the AI prelim and you only get 2 tries (3 if your advisor likes you a lot). So quite a bit of AI grad students never end up passing the prelim and have to change to another field or walk away with MS.

On a side note, is it general consensus that CMU is better for Machine Learning than Berkeley?

So, there is no one from MIT who can back up those well-known professors of Berkeley? I knew about MJ, SR, and JM from Berkeley. But, how about at MIT? Does anyone know any good ML professors? I know for CV, there are Bill Freeman, Antonio Torralba, and Eric Grimson; but not so familiar with ML professors.

[leogk]

I'm now a visitor at MIT vision group and finished MS degree at CMU.

The MIT vision group is one of best places... really friendly atmosphere and looks like everybody enjoys his/her life here. Another good thing here is that there have been a lot of very strong fresh PhDs as post-doc, so you may be helped out a lot by those guys. Personally, I was rejected in MIT and plan to go back to CMU, by the way... CMU is also good for CV, but the profs I'm interested in said the funding situation is very unclear, so may not hire a new student... (So, I'm now trying to apply for some fellowship programs these days... )

For ML, I think if you could be one of MJ's students, you'd better go for Berkeley. But in the fact that CMU has more than "20" ML profs, CMU may be a "safer" choice.

Tommi Jaakkola is the only guy I konw in MIT, but it's true that MIT machine learning group is not that big....

Above all those kinds of things, ML and CV are really competitive areas.... full of geniuses and extremely difficult to survive in academia after PhD graduation... Welcome to the Jungle !!

[fanta]

I'm now a visitor at MIT vision group and finished MS degree at CMU.

The MIT vision group is one of best places... really friendly atmosphere and looks like everybody enjoys his/her life here. Another good thing here is that there have been a lot of very strong fresh PhDs as post-doc, so you may be helped out a lot by those guys. Personally, I was rejected in MIT and plan to go back to CMU, by the way... CMU is also good for CV, but the profs I'm interested in said the funding situation is very unclear, so may not hire a new student... (So, I'm now trying to apply for some fellowship programs these days... )

For ML, I think if you could be one of MJ's students, you'd better go for Berkeley. But in the fact that CMU has more than "20" ML profs, CMU may be a "safer" choice.

Tommi Jaakkola is the only guy I konw in MIT, but it's true that MIT machine learning group is not that big....

Above all those kinds of things, ML and CV are really competitive areas.... full of geniuses and extremely difficult to survive in academia after PhD graduation... Welcome to the Jungle !!

Wow.. I see.. you are very experienced I suppose. I am thinking about co-advisors of JM and MJ at UCB (of course, there is no guarantee at this point...) but I am not sure who I can do this with at MIT. But, aren't Freeman's students tend to stay in academia well enough? I am not sure.. man this is such a hard decision.

By the way, congrats for going to CMU!

[fanta]

By the way, what all those MIT's AI department professors doing (if there is not many ML and CV)? Are they doing more pure AI AI?

[leogk]

Hey, fanta.

If I could have JM and MJ as my adviors, I wouldn't consider any other options..

You can see the details on CSAIL webpage, but it looks like there is a huge number of robotics people here.

The whole second floor of Stata center is assigned to robotics... it looks like a factory..

Anyway, I know it's gonna be one of the most important decisions in your life. Take time and think hard!

[bernard]

To decide, do a background check on every professor in the department, and find out where they got their PhD, and put their papers in Google Scholar to see how many cite them. I bet you will know their circle of friends once you searched a bit deeper.

[tedjj]

I thought Kanade is the most famous in CV. Which prof from Berkeley did you mean?

Probably he meant Professor Jitendra Malik. Kanade is famous, but I think he almost retired or did already.

Kanade is still active.

http://www.informatik.uni-trier.de/~ley ... Takeo.html

It is impressive:)