The conference proper started Tuesday morning at 8:30 with an opening ceremony and then an interesting invited talk by Leslie Pack Kaelbling encouraging a greater role for reasoning about and updating belief states as an underlying model for robotics. Unfortunately, I filtered out some of it because it inspired some ideas about probabilistic planning that I had to explore in my mind before I forgot them. Then I went to an IAAI invited talk by Jay Tenenbaum about his efforts to apply artificial intelligence to the field of cancer treatments. Essentially, he is building a very large database of how different treatments affect people, indexed by the subtype of their malignancy and, when available, the full genetic code of the tumor. This has already led to some surprising results where a drug that had been known to be effective in one rare subtype of melanoma was found to also be effective in a more common subtype of cancer of another organ. His premise is that cancers are highly individual, and thus that to find similar cases to a new patient we need to look beyond the experience of their individual physician. It sounds like a great and potentially achievable goal, though I shall have to discuss this with my wife (who is a resident in a family medicine program).
Then I went to a session on Search. The first two papers were well-presented and very accessible for someone who has thought about search problems but not done any serious research on them. The third may have been good, but the presenter spoke softly and did not carry a big stick, so I did not bother paying attention.
For lunch I met one of my co-authors at Goodfella’s Pizza across the street from the conference venue to tighten up the presentation of our work that I will be giving on Thursday. Our discussions were fruitful, but the service at the restaurant was positively awful. We both ordered sandwiches, and they were not ready for 45 minutes. Mine might have been a bit understandable as I ordered in the midst of a crowd of conference attendees that had mobbed the place at the same time. But he ordered after it had mostly cleared out, and was still waiting long after the restaurant was practically deserted. Did I mention that I ordered a fried chicken breast sandwich and got sliced turkey instead?
After lunch was a broad invited talk about General Game Playing. I did not find it particularly informative since I have already done some work in that area, but it did inspire me to try to find some time to do more with it. I spent the rest of the afternoon at sessions for the first AAAI Symposium on Educational Advances in Artificial Intelligence. I do hope *someday* to be offered a teaching job in which this sort of information might be useful. All of the presentations were interesting, but I was especially impressed by the first one I saw, about using projects based on the Pacman domain to teach a broad swath of AI techniques in a way that is fun and lets students implement algorithms without having to write or understand all of the boilerplate code around it. Given the opportunity, I will definitely be trying to use this.
At the end of the day there was a student reception scheduled. I was not in the mood for schmoozing with strangers (nor am I ever, really), so I showed up but only stayed long enough to verify that no one else was interested in taking the initiative to get to know me. There is a Hard Rock Cafe in that area. I had kind of wanted to go there since I have never been to one and I am, in fact a fan of hard rock. I assumed it would be too noisy for the kind of meal where you want to have serious conversation with your dinner-mates, so this seemed an ideal time to try it. I think they should rename themselves to the “Popular Music Cafe”, as their music and memorabilia span all types of modern music. The music (which contrary to my expectations, can barely be heard above the clatter and chatter of diners) came from concert footage and music videos shown on various screens. I saw many genres of music: funk (Kool & The Gang), rock & roll (Bruce Springsteen), pop (Billy Joel), but only one song by a hard rock band, and perhaps their song that rocks the least hard (“Why Can’t This Be Love”). The memorabilia hanging on the wall included pieces from such hard rockers as Sheryl Crow, John Lennon, Willie Nelson, and even Ella Fitzgerald and Branford Marsalis. So yeah … The food was good.
Allow me to explain another travel annoyance, how MARTA managed to weasel an extra $0.50 from me. (The horror!) The system for accepting payment for transit is that you buy either a plastic card for $5.00 or a cardboard ticket for $0.50, either of which has an RFID tag (I presume) that is read when you tap it at a subway or bus entrance. These cards/tickets are purchased from vending machines, at which you are also able to add value of various types to the card/ticket — a single trip fare, a daily unlimited ride pass, a certain amount of cash, etc. Nothing unusual so far, except that you have to pay for the media. Although it is not a bargain if most days you will be making a single round trip, I desired to purchase a weekly unlimited ride pass for convenience. The vending machine only offered 1-4 day passes, so I bought a 4 day and assumed I would reload more value onto it later.
Tuesday night on my way home I tried to reload my ticket, since my 4 days would be ending at whatever they counted as the end of the day. The vending machine happily accepted my request to reload value onto my ticket and displayed a menu of the various types of value that can be added. However, none of the buttons for any of these menu selections did anything. I asked a MARTA employee who was emptying trash cans if he knew anything about the vending machines, and he tried the same things I had, then told me that tickets cannot be reloaded. Although this was obviously not true, I consented to purchase a second ticket for the next two days.
Wednesday morning I woke up to some frightening news: my advisor had sent an email at 3:00 saying he was in the emergency room. I called some hospitals and found the one he was at, where they promised to have him call me. (Thankfully, I suppose HIPAA does not cover whether or not someone is a patient, or there is an exception for ongoing emergency care.) An hour later I found out that he had experienced difficulty breathing, was having tests done, and did not want me to come in.
I had missed the plenary session, but went to the last half day of EAAI. The first part of this was a talk on incorporating active learning. I had already seen something very similar in the Lehigh Teacher Development Series, but it was a good reminder. We then split into groups to put this into practice devising strategies for teaching different topics in AI. Glancing through the names of the people in my group, I realized that one of them was a faculty member with whom I had had a telephone interview at Washington & Jefferson college. I tried to make a joke about their having not hired me, but I fear she may have interpreted it as genuine bitterness. I suppose it does not really matter. After this was a panel session among experienced educators, which was also somewhat useful.
I found some familiar faces with whom to eat lunch, then saw an invited talk about designing agent interface mechanisms (such as auctions and voting) in such a way that no agent can gain an advantage by misrepresenting their true desires. I lacked some necessary background, but the problem and examples at least were interesting. Then was the first AAAI session that was actually about topics peripherally related to my work. The first two talks were good, but I did not hear much of the third one because I received a voicemail from my advisor’s wife saying that they were keeping him overnight and that he had a favor to ask me.
I tried calling his room in the hospital for 40 minutes, getting either busy signals as he talked with others or endless ringing as he was being seen by the staff, then decided to just go to the hospital and hope they would let me in to see him. They did, and from my perspective the nurses and other staff at Emory University Hospital Midtown were really exceptionally helpful and friendly.
The diagnosis was pneumonia. I did not get a report on the etiology, but the antibiotics they had been giving him seemed to be working. They were holding him in 23-hour observation with intentions to release him at some point the next day unless he took a turn for the worse. We visited for a few hours with another colleague and friend who planned out how to divide the responsibilities that Dr. Munoz-Avila had had at ICCBR-10 the following week (and would not be attending, to try to rest and recover). About 8:00pm I left to give him a chance to try at sleeping, and ended up doing just that myself quite early.
Thursday morning I skipped the plenary invited talk to instead check how my slides looked on a projector, and found that the figures were virtually illegible from the back of the room. After adjusting them appropriately I went to an IAAI invited talk by science fiction author Vernor Vinge. He talked about the technological singularity as an emergence of intelligence so superior to that of current humans that it can only be compared to the separation of Homo Sapiens from the rest of the animal kingdom. I am rather a skeptic on this topic, but he was interesting to listen to.
Then I went to the first session on planning, where there were two good talks and one that I did not really pay attention to. The group that I joined for lunch wanted to visit a Vietnamese noodle restaurant, so I bravely tried Pho. It was edible, but I have no desire to eat it again. The soup contains very long noodles that are nearly impossible to collect on a spoon. I tried imitating the use of chopsticks by my fellow diners, but that was mostly a comedy of errors.
The second planning session was after lunch, and I again greatly enjoyed the first two talks and struggled to focus on the third. Then there was a session specifically on planning under uncertainty. These were somewhat further afield from my ability to really understand what is going on, but still interesting. I got an email from my advisor that he had been released and was at the airport ready to head home.
The very final session of the day was the one in which I would be presenting. Although the first talk in it was somewhat planning related, it seemed to be a mixture of whatever was left after filling the more focused sessions. My presentation went ok, though midway through I forgot how I intended to use an example and had to move on without explaining it. The questions afterward were insightful and I was mostly able to answer them, and no one was jumping up and down saying that what we had done was stupid or had been studied 20 years earlier.
I rested a bit in my room before packing up, but got to the airport around 8:30pm. My flight would not be leaving until 6:00am, but I would not be able to get enough sleep overnight for it to seem worthwhile paying to stay in my room another night. I planned to check in and doze / read / game at the gate through the night.
There was no one at the U. S. Airways check-in desk, so I started to use one of their self-check-in terminals. After it determined who I was, it informed me that I would not be allowed to check-in more than 8 hours before my flight time. Annoying, but I could sit on the floor and read for an hour and a half. A few minutes after 10:00 there were still no ticket agents, so I attempted to use an automated terminal again. At some point in the meantime, however, they had all been shut down.
I asked a passing security officer if there was ever anyone at the ticket counter, and he said only from around 4:00am until 6:00pm. So rather than resting comfortably at the gate, I spent the night sitting outside the security zone, watching my luggage to make sure no one walked off with it. How can the busiest airport in the country be effectively closed down for 10 hours each day? I was *not* happy. I’ll let you guess whether or not the Atlanta airport has free wireless Internet access available.
I found a seat on the floor near a power outlet and kept myself awake playing a game of Civilization IV until 4:00am, when I checked in and got to my gate around 4:40am. I tried sleeping here, but that did not really work. I did doze through most of the uncomfortable flight. Back in Philadelphia I caught up on some more reading and wrote most of this entry while waiting 2 hours for my flight back to Williamsport.
The final leg of my journey was on a prop plane, which was a first for me. I was expecting to find the experience somewhat different from those on jet-powered vehicles, but this was not really the case. Sadly, I did not have the opportunity to fight a Nazi while dancing around the spinning craft’s propellers. Home, sweet home! I got back to Williamsport around 1:00pm, and slept for all but about 3 hours until 8:00am the next morning. I have never felt so refreshed in my life. My only regret is that through 6 days in Georgia I did not see a single peach or peach-based dish for sale, and so I never ate one.