Jon Udell: Working with Bayesian categorizers

Jon Udell of InfoWorld and the O’Reilly Network is playing around with Bayesian classifiers for blog-post categorization. He didn’t have much luck in his first tries, but he also didn’t have many training examples. He’s concluded, quite rightly, that he needs a nice interface for sorting examples and training classifiers.

I have a different idea. Why not let the algorithm come up with the categories, not just the categorizations? I would start with a hierarchical, incremental algorithm like COBWEB, and build an interface with two parts: (1) a means for easilly handing a document to the system for categorization, and (2) a category tree browser, allowing easy browsing and viewing of the documents. This could be a part of someone’s own personal Google.

Read the rest of this entry »

FedEx Institute of Technology

Wired News has an article on the new FedEx Institute of Technology at the University of Memphis. The emergence of a new corporate research lab that does AI research (among other things) is exciting, hopefully it’s a sign of better job prospects for new CS Ph.D. in the near future.

I only wish there were more real scientific content on their website. As it is right now, it’s frustratingly devoid of any specifics about their research. Neither the Institute’s researchers nor their publications are anywhere to be found on the site, and the pages describe the research in only the vaguest terms. This could be the kind of place I might want to appy for a job when I graduate, but how would I know? I can’t tell what they do.

Read the rest of this entry »

Forum for AI, Nov 21

At this Friday’s Forum for AI will be Peter Stone, assistant professor in the UT Artificial Intelligence Lab, speaking on The RoboCup Challenge: Progress and Research Results in Robot Soccer. Peter was one of the main people behind Carnegie Mellon’s CMUnited robot soccer team, that was a multi-year RoboCup world champion. Now he’s in charge of UT Austin-Villa, our local RoboCup team.

Note: This talk will be at 3 p.m. in ACES 2.402

Read the rest of this entry »

Nov 7 Forum for AI

Tomorrow’s Forum for AI will feature Bill Smart from Washington University in St. Louis, talking about value function approximation in reinforcement learning. I’m excited about this talk, since my research area is RL in continuous state spaces (i.e. the real world) and value function approximation is pretty much necessary to do RL in the real world. I’m particularly curious about moves toward convergence proofs for VFA. Look for a summary here over the weekend.

The talk will be at 10 a.m. in ACES 2.402. In case you can’t tell from the time of the talk, this won’t be a brown-bag lunch talk.

Read the rest of this entry »

Posted in AI, Science. 1 Comment »

U.S. to Resume Manned Moon Program?

According to Space Daily and Bush may be planning to announce a new space policy focusing on returning men to the Moon and developing technology for manned flights to near-Earth asteroids and Mars.

If true, this is great news! I find the West’s risk averse, spendthrift attitude toward new technology and exploration depressing. I want to look up at the Moon and think: There are people up there!

If that is the plan, I wonder what role of the Space Shuttle will be. The reusable shuttle was supposed to make spaceflight cheaper, and instead it made it much more expensive, sucking up budgets that could have been spent better elsewhere, and anchoring us firmly and disappointingly in low-earth-orbit.

If Bush wants to be really bold on the Kitty Hawk centennial, he’ll say that the plan is to retire the Shuttles immediately, or within 2 years, and
develop disposable capsules (none of this space plane crap) that can serve as transport to the space station and to high orbit and beyond.

Read the rest of this entry »

Lakoff Starts a Progressive Think Tank

I’ve already mentioned George Lakoff and his applications of the cognitive science of language and metaphor to current political thought. He mentioned that the reason that conservatives were winning the political struggles in this country was that they have better metaphors — they’re better a framing the debate. Then hoped that he would give some constructive help. Looks like he’s doing just that. He and some other UC Berkeley and UC Davis faculty have
started a progressive think tank
to work on how progressives can use language to better frame the issues. The article is an interview with Lakoff, and the deepest quote is on the second page contrasting the conservative approach of investment in cognitive infrastructure to the liberal approach he says:

Also, within traditional liberalism you have a history of rational thought that was born out of the Enlightenment: all meanings should be literal, and everything should follow logically. So if you just tell people the facts, that should be enough — the truth shall set you free. All people are fully rational, so if you tell them the truth, they should reach the right conclusions. That, of course, has been a disaster.

Though I think an equally plausible theory is that Democrats held power in Congress for so long that they became complacent, while the republicans had to struggle to find a way to reach people.

Read the rest of this entry »

Forum for AI: Tony Cassandra POMDP redux

Tony Cassandra gave a nice talk Friday for FAI. I had had an informal intro to POMDPs earlier (from Tony), but it was nice to see it presented more formally. The issue of how to reason and act with hidden state (i.e. in “partially observable” domains) is going to be very important if we want intelligent agents that deal autonomously with people and other agents, since by default the other agents’ internal stage (what we’re thinking) is hidden.

One important thing is that all the work that Tony presented, and (I think) all the current work on POMDPs assumes that you have a (probabilistic) model of the environment. How an agent can learn such a model is still totally an open question. Agents that can build and/or adapt their world model on-line are a great field of research (if I do say so myself. ;-) )

Apart from the great content, Tony’s talk was interesting on a couple of other levels. First, instead of using powerpoint, or any kind of slides or “slideware”, he just did the visuals for his talk as one big web page. This is interesting because the day before the talk I received my copy of Edward Tufte‘s essay The Cognitive Style of Powerpoint, which I’ve mentioned before. Tufte describes in great detail the ways in which PowerPoint is bad for communicating complex ideas. Particularly bullet lists of short sentence fragments and breaking up the presentation into slides. Tony’s talk web page has breaks in natural places for the subject matter, and he just scrolled down as needed while talking. He also has (mostly) full sentences describing ideas and few bullet lists.

The other reason it was interesting was that to present the talk he just put up the web page on his server. Then when he arrived he used my laptop connected to the CS dept wireless net. Not only did everything work flawlessly, but I never had any doubt that it would. How’s that for ubiquitous computing?

Read the rest of this entry »