What sort of good facilities/labs/mentors are there?
If I wanted to study AI there, should I go for CSC instead of ECE? Not that I think it matters too much. I assume it depends on me.

2/11/2013 2:55:28 AM

There are several excellent professors at NCSU who focus on AI. If an undergrad, talk to the CSC undergraduate advisor to get a full list of names and visit them during their office hours. Or, if a graduate student / postdoc talk to the CSC graduate school director to determine who is looking to take on new folks.

2/11/2013 8:43:37 AM

my ex is getting a phd in AI assisted learning at ncsu and it's squarely in CSC.

2/11/2013 10:14:54 AM

It all depends on what you would like to do. AI is a really broad term that can mean a lot of different things. My advisor, Dr. Doyle, is one of the AI professors here and I am familiar with most of the others.

Do you have specific interests? It might help you to ask yourself what smaller problems within AI you would like to be involved with. You are not going to work on some sort of Kurzweilian general AI for your graduate work.

You should not go into ECE if you want to do AI , especially at NCSU. I looked into this option as well. If you are interested in robotics, I would go to CSC and get an ECE professor to be on your committee.

When I came out of undergrad I also wanted to do AI and it's been a long journey since. Feel free to ask questions or shoot me a PM.

2/11/2013 10:16:24 AM

Yeah AI is actually a very broad category. There are guys who work in AI that's much more Linguistics or Learning based (Lester) or Human Factors (St. Amant) vs the more "pure" AI guys.

Actually back when I shared some lab space with some of St. Amants guys there were one or two guys who didn't have CSC undergrads who were working on some simple robotics AI. They were doing the software side on prebuilt stuff but you'd be surprised how many AI guys don't come from a pure CSC background.

2/11/2013 10:41:11 AM

There are ECE AI teachers. They are not good.

2/11/2013 11:55:50 AM

Have you looked into operations research?

2/12/2013 11:21:23 AM

Well I do want to work on AGI but more specifically I want to find ways to make analogies between fairly different topics. Also data compression/storage/recall are some things I have ideas for. NLP is something else I could easily get into. Anything involving pattern recognition and knowledge representation really.

2/13/2013 10:48:47 PM

Quote :

"Well I do want to work on AGI but more specifically I want to find ways to make analogies between fairly different topics. Also data compression/storage/recall are some things I have ideas for. NLP is something else I could easily get into. Anything involving pattern recognition and knowledge representation really."

Analogies? Sounds like you've been reading some Hofstadter? Reading GEB while taking CSC 420 here at NCSU is what got me interested.

To my knowledge AGI/Strong AI does not get researched that much in a university setting (in fact, the term AI is itself a little out of vogue now, because it comes with a lot of baggage). Why doesn't AGI get a lot of university attention? Because you have to secure funding from agencies to do it and, given some of the failures in the 80s along with agency budget cuts, those types of proposals are less likely to get funded.

However, all of the subcomponents that may be involved in that type of research program do get funded - deep imaging of the brain, computational neuroscience, machine learning, computational biology, knowledge representation, NLP, etc. That's why I suggested you pick an area you would be interested in working on. I would suggest you look through the list of professors here and see if you like any of their research projects:

http://www.csc.ncsu.edu/research/areas/index1.php#3

You should do this for any school you are considering. Finding a compatible advisor is one of the biggest keys to success and being happy in graduate school.

Hope this helps.

2/14/2013 9:24:44 AM

Yeah hate to break your bubble but we're nowhere close to getting any true "sci-fi" AI. It might be a noble endeavor to do that research but from a practicality standpoint think about the end product you might be interested in working on and work your way back.

For instance, if you're interested in talking to a computer and having it do something go NLP.

Young and Lester do a lot of work in that area or at least did when I was in school.

2/14/2013 10:38:02 AM

Yeah I realize how far off the real thing is, but that is all the more reason to start now.

Quote :

"Analogies? Sounds like you've been reading some Hofstadter? Reading GEB while taking CSC 420 here at NCSU is what got me interested. "

I think he understands human thought better than most people in his field these days. He did a talk at Stanford called "Analogy is the core of cognition" which was brilliant imo. It is on iTunesU if you want to check it out. If he is right, then creating a truly intelligent mind might actually be possible with just a few desktop PCs. Or at least it might be possible to build the technology yourself without having to invest hundreds of thousands of dollars. Also, still assuming he is correct, methods of data compression and pattern recognition based on his concepts would be incredibly robust and fairly immune to noise/corruption. It is definitely something I'd like to explore, if I can find something concrete to work with.

2/14/2013 1:08:20 PM

Quote :

"Finding a compatible advisor is one of the biggest keys to success and being happy in graduate school. "

QFT

2/14/2013 1:09:18 PM

^ lol yeah just remember that unless you're doing a PhD the program is too massive for your adviser to really give a shit about your Masters work or at least that was what I saw usually.

That said I liked my adviser and it was kinda an unspoken thing that I just was there to just to get the piece of paper and salary bump. I wasn't getting funding or anything so it was always just do the work and push on through which both sides were fine with.

Probably a different story if you're getting money but that's pretty much reserved for PhD's since the program is like 400 students grad school.

2/14/2013 1:15:03 PM

Quote :

"I think he understands human thought better than most people in his field these days. He did a talk at Stanford called "Analogy is the core of cognition" which was brilliant imo. It is on iTunesU if you want to check it out. "

Seen it, read it, listened to it in addition to all his other works - I love Hofstadter. GEB, I am a strange loop, Metamagical Themas, and The Mind's I have had a profound impact on my life.

What I have come to realize is that Hofstadter isn't a researcher in the traditional sense. In the introduction to the 25th anniversary edition of GEB he admits as much. He says something like winning the Pulitzer for GEB so early in his career freed him up from the publish or perish culture. He has a handful of papers published in journals, which by most standards for academics, is very low. If you are planning on being an academic, you will find life to be very different. You're ability to publish in good journals and secure research grants will determine your ability to maintain a career. To my knowledge Hofstadter has managed to escape all of this, which is very rare.

I love him as a writer and as a popularizer of meaningful ideas, but I don't think he is taken very seriously by most researchers in the field. I don't mean that as a slam, I just don't think that is what he does.

Another point worth considering is most of what Hofstadter does is more properly called cognitive science. It is true that there is a good deal of cross over between cognitive science and AI, but Hofstadter's primary appointment is in cognitive science, which is a branch of psychology. It's mainly semantics, but something worth considering if you are trying to decide which area to pursue. Hofstadter is still taking graduate students at Indiana as far as I know.

Anyway, wildly extrapolating from your few posts in this thread, you remind me of myself about 4-5 years ago. I ended up taking a somewhat weird path but am very happy with where I am now. I ended up getting a MS in statistics along the way and now mainly I work on problems in genetics and personalized medicine but using approaches from AI. My ideal post-doc would be doing computational neuroscience, so I still feel like I'm in the area I set out to be in. This is what I mean by picking problems from subfields of AI to get your foot in the door. You have to carve out a niche in your field before you can start working on bigger projects.

From your interests though, my advisor might actually be a good fit for you. This is his book (which is totally different stuff from what I do):

http://www.csc.ncsu.edu/faculty/doyle/em2m/

Here is one of his papers:

http://www.csc.ncsu.edu/faculty/doyle/publications/wirp83.pdf

You can look through his publications and see if there is anything else that interests you:
http://www.csc.ncsu.edu/faculty/doyle/publications/



[Edited on February 14, 2013 at 1:28 PM. Reason : ]

[Edited on February 14, 2013 at 1:43 PM. Reason : papers]

2/14/2013 1:27:11 PM

Oh yeah I'm totally aware of Hofstadter's status in academia. He is well respected but to the more serious scientist his ideas are more like recreational thinking. Which I have no problem with. I just see a lot of value in his opinions because he isn't directly associated with AI research, which means his perspective on the subject isn't as prone to tunnel vision. He can be more creative without having to worry about his reputation. Sometimes I think it's healthy to be mentally flexible. But then of course it can be a slippery slope into PSEUDO-SCIENCE if you're not careful. Next thing you know you're publishing books about quantum consciousness and the morphogenetic field.

2/14/2013 4:37:19 PM

Bump for this Atlantic article:

http://www.theatlantic.com/magazine/archive/2013/11/the-man-who-would-teach-machines-to-think/309529/

11/2/2013 9:26:05 PM