Artificial Intelligence or High Performance Computing?

[anon1]

Wondering which is a better grad school focus.

AI algorithms sound like more fun, but is AI less practical than HPC?

HPC sounds more useful/employable, but also more frustrating/boring.

Any advice? I'm willing to study something frustrating and possibly boring (though maybe I'll love it once I start working on it) if it will lead to good job prospects.

[R Deckard]

There is no reason you can't do both.

[dawidbrook]

Basically, I Prefer going for Artificial Intelligence because of it is design for the task that are not being performed by humans. In super computer the commands are to be given by human while in Artificial Intelligence it is preloaded. that's way it is more useful and smarter.

[anacron]

Wondering which is a better grad school focus.

AI algorithms sound like more fun, but is AI less practical than HPC?

HPC sounds more useful/employable, but also more frustrating/boring.

Any advice? I'm willing to study something frustrating and possibly boring (though maybe I'll love it once I start working on it) if it will lead to good job prospects.

Don't limit yourself, research shouldn't be just about one sub-field or another. You should really aim for what interests you and where your research takes you. I'm actually in the HPC field but I'm moving towards applications of machine learning algorithms and molecular dynamics algorithms.

So, I wouldn't go into grad. school thinking that its one or the other. It should really be an opportunity for you to explore whatever interests you.

[anon1]

Even if I have an adviser at each school in each field who wants to work with me... ? Just don't know how easy it is to change advisers, or how expected.

Also, if you don't mine my asking, if you had to choose between UCLA and UCSD, which would you pick...?

[Azazel]

HPC sounds more useful/employable, but also more frustrating/boring.

If something sounds boring, don't study it for your PhD

And AI can be very practical. Just ask the ML people.

Even if I have an adviser at each school in each field who wants to work with me... ? Just don't know how easy it is to change advisers, or how expected.

Also, if you don't mine my asking, if you had to choose between UCLA and UCSD, which would you pick...?

I'd probably choose UCSD. But this should be a very personal decision, based primarily upon advisor fit. How well do you like the potential advisors at each school? How well does the research they do fit your interests?

[anacron]

Even if I have an adviser at each school in each field who wants to work with me... ?

Actually, this makes it much easier. You can always do one project for the adviser in HPC and another project for the adviser in AI. Its not uncommon for young grad. students to try different fields before "settling down". If you make your situation clear to each adviser and they are open to this idea, then you shouldn't have too much to worry about.

Its not until you get your hands dirty that you truly realize what you're interested it - this certainly happened to me as I transitioned from undergrad to phd...

Also, it's completely possible to be co-advised during you PhD. So, if you had an interdisciplinary research topic then there shouldn't be any issues with working with both Profs. Of course, you need to do the hard work of coming up with an interesting project and convincing the two profs - but its doable and people have done it before.

Also, if you don't mine my asking, if you had to choose between UCLA and UCSD, which would you pick...?

I'm not sure I can help with this... I don't know many faculty from LA or SD. However, San Diego has a supercomputing facility nearby so it may be marginally better at HPC (can't say for sure though).

If I were in your situation, I would talk to the AI and HPC profs at both schools about:

how flexible they are to changing advisers (in the event that you become disinterested in one of the sub-fields),

how easy it will be to collaborate between HPC and AI.

[anon1]

Azezel & Anacron-- the professor who wants to work with me at UCSD does HPC. He was super nice, but idk that I def. want to do HPC. I've never even taken one class in it, and I'm more interested in theory, which seems closer to AI. I was a bit intimidated on both my school visits, so the fact that this prof was nice and encouraging, and kept saying that he really wanted me to go to UCSD made me like him more, though idk if this is a good way to choose between schools or not (since after I get there I may feel comfortable and realize I want to do something else). The UCLA prof does theoretical/game-based/problem-solving AI stuff. I've also never taken an AI class though. The AI prof was more stern, not as nice, and one of his students said that he was unhelpful as an adviser--- though his other student said he was the most amazing prof ever, and another random student was working his tail off in this professor's class just so he could have a shot at being in his research group. (he also got amazing reviews on rate my professor, though i know this isn't necessarily a good way to judge a research adviser)

Though, it sounds like I can change advisers when I get there, so maybe I should be looking at the faculty as a whole, instead of the two profs who will be my initial advisers.... I know I'm interested in theory, and UCSD has a better theory program, but if I go to UCLA I'll be essentially starting out with a theory person... it's hard.

Also, within CS do you think UCSD is recognized as a better school than UCLA? or visa versa? US News has the two tied at #14, QS world CS rankings have UCLA at #12 and UCSD at #33, though UCSD does way better (even better than CMU and other top schools) in the 'citations' category, and the National Research Council Rankings put UCSD ahead in research, student outcomes, and diversity, but UCLA ahead in two vague categories that are something like "what scholars deem a good school should have"....

UCLA seems like it would give me the overall prestige factor, right? But I get the impression that UCSD is better for CS and might give me a better overall CS education....

Is one location (La Jolla/San Diego VS Westwood/Santa Monica/LA) better in your mind...?

A final option is to reapply again this year, because I am so torn, and because I really did not do the best job at 1.pinpointing what it is I want to study, and 2. searching for faculty members who are doing that kind of research... I put together an application that was not my best (after doing internships at Harvard, MIT, and now NASA), so I also wonder if I might have a shot at a higher ranked program (I know this is superficial, but it's that what-if that bothers me).

Anyway, thanks again for your thoughts.

[Azazel]

Azezel & Anacron-- the professor who wants to work with me at UCSD does HPC. He was super nice, but idk that I def. want to do HPC. I've never even taken one class in it, and I'm more interested in theory, which seems closer to AI. I was a bit intimidated on both my school visits, so the fact that this prof was nice and encouraging, and kept saying that he really wanted me to go to UCSD made me like him more, though idk if this is a good way to choose between schools or not (since after I get there I may feel comfortable and realize I want to do something else). The UCLA prof does theoretical/game-based/problem-solving AI stuff. I've also never taken an AI class though. The AI prof was more stern, not as nice, and one of his students said that he was unhelpful as an adviser--- though his other student said he was the most amazing prof ever, and another random student was working his tail off in this professor's class just so he could have a shot at being in his research group. (he also got amazing reviews on rate my professor, though i know this isn't necessarily a good way to judge a research adviser)

Though, it sounds like I can change advisers when I get there, so maybe I should be looking at the faculty as a whole, instead of the two profs who will be my initial advisers.... I know I'm interested in theory, and UCSD has a better theory program, but if I go to UCLA I'll be essentially starting out with a theory person... it's hard.

Also, within CS do you think UCSD is recognized as a better school than UCLA? or visa versa? US News has the two tied at #14, QS world CS rankings have UCLA at #12 and UCSD at #33, though UCSD does way better (even better than CMU and other top schools) in the 'citations' category, and the National Research Council Rankings put UCSD ahead in research, student outcomes, and diversity, but UCLA ahead in two vague categories that are something like "what scholars deem a good school should have"....

UCLA seems like it would give me the overall prestige factor, right? But I get the impression that UCSD is better for CS and might give me a better overall CS education....

Is one location (La Jolla/San Diego VS Westwood/Santa Monica/LA) better in your mind...?

A final option is to reapply again this year, because I am so torn, and because I really did not do the best job at 1.pinpointing what it is I want to study, and 2. searching for faculty members who are doing that kind of research... I put together an application that was not my best (after doing internships at Harvard, MIT, and now NASA), so I also wonder if I might have a shot at a higher ranked program (I know this is superficial, but it's that what-if that bothers me).

Anyway, thanks again for your thoughts.

If you're really not sure what you're interested in, I'd consider a few things:

- How interesting do you find the general faculty at each school? (e.g. which school is better if you want to switch advisors)

- Which advisor is most (or at all) amenable to "letting" you switch, if that's what you want? While no advisor can stop you outright, some can probably make it painful. (Sounds like UCSD has the edge here).

- Which advisor is most open to interdisciplinary collaborations? (e.g. what anacron mentioned) Do they expect you to come in and work on a big project, or are they open to something new? You can try to figure this out via email, if you're careful.

- What are common student outcomes for each advisor? How well are students placed, and how well are they doing?

Depending on your answers to these questions, it may be better to reapply, although UCLA and UCSD are both good schools. If you reapply, will you be getting any new (e.g. better) letters of recommendation this time around? Rec letters are often the determining factor, especially at top schools.

And I think the difference in prestige between UCLA and UCSD is pretty much a non-issue. While I've never lived in LA or San Diego, I'd probably lean toward San Diego (better weather, less traffic/sprawl, more laid back).

Edited July 29, 2012 by Azazel

[anon1]

Actually, the UCSD prof who does HPC is highly interdisciplinary; he works with scientists at UCSD and at national labs, under large grants. His students write software packages and do parallelization to support scientific research. Since he gets large grants I would get a few thousand more per year above the regular stipend and would only TA for 1 quarter.

At UCLA I would TA for every semester after the first, if I stay with the same adviser. His work is too theoretical to garner government grants and it doesn't seem to be interdisciplinary.

2 of my rec letters were solid, but 2 were so-so. I have another person I could ask (I think) so that might help my chances. I have an internship at NASA right now, but after that I don't know what I would do. Do you think taking time off would look bad? Maybe I will try e-mailing some of the professors I am most interested in working with to see if they think I would have a shot at getting in...

I've looked at jobs of recent grads and it seems that the HPC person's students go for software engineering jobs at high profile companies, or else are working at national labs, while the AI professor's students are more inclined to stay in academia, or do research: one at NASA, though some do software engineering too. Although the AI prof's website listed lots of people from the 90s, so it may be that the HPC person has had more recent grads or else gets them through the program faster...

I think UCSD's higher concentration of theory faculty makes UCSD have the edge for alternative advisers. As far as switching goes, I think either would let me.

Anyway, thanks again for your thoughtful advice, I appreciate it!