Tuesday, May 5, 2009

That didn't take long

Remember a few posts back when I was cynically forecasting the doom of free speech in America with regards to posting whatever you want to a blog (or really any other electronic media)?

Well, sorry, to say, such things seem to be coming to pass. Granted what's below is only a proposed law, and it seems extremely unlikely to pass (and even more unlikely to withstand its first test of constitutionality), its still pretty appalling that such things are being proposed in America, where we should expect out elected representatives to be at least familiar with the constitution.

Weekend Update

This weekend: I studied data mining for twelve hours, discovered that Quake Live went live a while back, took the GRE, wrote a pretty interesting research proposal for data mining, went to a dance party at North Side Tavern, bought a Pizza Pan pizza from a couple of guys standing outside Fries Cafe, saw an awesome performance at the CAC for 44, went to a Taste of India event, and jammed very productively with Matt.

So, more detail. I've been struggling with the book that Dr. Yu assigned for data mining. I'm learning from it, but its written at a very theoretical level and I have a gap in trying to implement it in R. I found a book online thats at a more applied level, and going through its case studies has been very helpful: .

Quake has long been a favorite vice of mine, but I havn't played in probably five years. They've just implemented an online version of the game; pretty awesome in that it runs in a browser with better graphics than the original Quake 3 had. Fun times, but gotta keep myself from playing it too much.

I took the GRE Sunday afternoon. I figured that I need to have a baseline to compare myself against and I would like to be able to have a score that I can use on applications that request them. Much as it would be wise to study real hard for it and try to maximize my score, I figure I have plenty of other very interesting (and difficult) things that I would have to forego if I were to concentrate on that. Plus I feel like its the honest thing to do to represent myself at my native, uninflated level of capability. With that said, I got a 660 on the verbal and 570 on the quantitative. Not top tier, I realize, but significantly better than I did on the gmat (in terms of quant). I looked up UCSD's GRE stats for cognitive science (just for the sake of comparison)... apparently their 2005 class had something like 680 verbal and 760 quantitative. 760 is incredible. I looked up the GRE distributions by declared feild of study, and 760 is above the mean even for people who are studying mathematics at the graduate level. Amazing. Well, every program I've looked into says that they don't have a floor score for admission; that they take the full profile of the applicant into consideration. I'll have to develop a very attractive overall profile, eh?

For data mining I'm doing a project with census data that attempts to determine the demographic factors that best predict economic success. I included some other incisive questions touching on the social cost of affluence: does economic success correlate to other positive measures of lifestyle? The results will be interesting.

Saturday night I went to Dance MF; the acronym standing for the colorful euphemism you might guess. The played a great mix of music and there was an active population of uninhibited dancers moving in all sorts of fun and creative ways. Not sexual ways, interestingly enough, just fun and creative. I wore my cowboy boots, and I danced blisters into my feet. They played a mix of Zero (the Smashing Pumpkins song)... I was super excited for that.

On the way home, passing fries cafe, some guy was standing by the side of the rode with a stack of boxes waving and yelling at cars. The following conversation ensued:
Me: "Are those guys selling pizza!?"
Rachael: "Its just some drunk guys yelling at cars, forget it"
Me: "No, I think they had pizzas! Are you hungry? I could use a pizza"
Rachael: "I really don't think thats what's going on..."
Me: "I'm gonna go ask. He'll probably be like 'no, I'm just some drunk guy yelling at cars'"
Rachael: "..."

Turns out, to our (eventual) mutual delight, they were indeed selling pizzas. Five bucks is probably too much for a small Pizza Pan pizza (considering they'll give you three large ones for ten more dollars), but the comedy made it worth the premium.

Stepping back chronologically, Saturday evening I left work a bit early to go see a show at 44. "Science and Garfunkel" and "Fists of Love" played, both were awesome. The first was a solo act with a beautiful mustache. He had tons of electronic equipment, and as he was setting up I was wondering what he could possibly be using it all for, and if it was worth it. Turns out, it definately was. He had a boomerang, and he did some really creative things with it, creating a whole song structure on the fly with simple sounds, then doing really good guitar riffs with it. It was neat to see experimentation and creation happen live, right in front of me. Very cool. Fists of Love were also excellent, the last few songs in their set really grabbed me. They said that they're recording soon, so I'll be excited to get a copy of their stuff.

Sunday afternoon after the GRE I randomly saw a "Taste of India" sign for an event that was just about to happen. I love Indian food, so I went and called up Matt to join me. He and Jude and I had some delicious things I don't know the names of, saw some kids do some traditional Indian dances, then went back to Matt and Jude's place to jam. We did some blues stuff and riffed off of each other better than we ever have in the past, I think. We're building up to something good!

So yeah, it was a pretty fulfilling weekend.

Programming and Neural Networks

My Fundamentals of Programming class has changed in an interesting direction. The former professor has stepped aside and the head of the Biomedical Engineering department has taken over. This is exciting for several reasons: first, he's promised to post all of the lab assignments online so that those who are ahead of the curve can do them all at once and be done with the class. Second, he's one of the professors that Dr. Bickle recommended to me as good contacts for learning more about cognitive science, and he studies reinforcement learning. Third, (and related) he mentioned a project he worked on where they made a program that learned to play tic-tac-toe by playing against humans. It started out making random moves, but eventually learned the game well-enough to play against humans to a draw every time.

I sent him the following questions:
"On the tic-tac-toe game you mentioned, was the learning done with artificial neural networks? Could it be done with just nine input nodes and nine output nodes (mapped to the game board) and some larger number of hidden nodes? I imagine that using such a setup and reinforcement learning (with back-propagation(?)), the program would come to store every possible game state and would play a human player to a draw by recalling the stored "best" move from memory for each situation. Did the program you mentioned take this approach?"

I think that this is within my reach in terms of programming abilities. It would be fun to try to make it work, maybe with my minesweeper game.

If you have MATLAB, you can play the game by downloading the M File here. Even if you don't have MATLAB, you can download it and admire the code in a text editor. I'm trying to figure out how to compile it to make it run as a standalone program, but it seems unlikely that anyone would want to download an EXE from my blog anyway.