On Friday I received an email congratulating me on my paper's acceptance into DePaul's SORCS 2011 conference. It's a small student research conference, but it's still pretty cool to get something that I was the sole author on published. The paper is the one I mentioned earlier about using genetic algorithms for the purpose of autonomous music composition. It's based on the MIDarwin project I built for my machine learning course a while ago.
Quick Overview: MIDarwin takes in a MIDI file as an example of a "good" song. I found a whole library of MIDI files of popular songs to use for this and a lot of my testing used MIDI versions of Dammit and Panama. Then the program builds a population of randomly generated songs and compares them to the example song. The songs in the population that are most similar to the example song are deemed the "fittest" and are mated with other fit songs to produce another generation of songs (there's also some survival of good songs and some random mutation). As the population evolves the best songs in the population start getting more and more like the example song with the goal of producing music influenced by an already deemed "good" piece of music. The whole process is based off of Darwinian evolution as observed in biology.
I still need to make some revisions based on the reviewer's comments, but I'm pretty exited (and a little nervous) about presenting by my self at a conference. I need to prepare a 20 minute talk, plus slides about my research. Fortunately I know the stuff pretty well ( I did write a paper on it... ), but I still need to get everything together and I'd like to make some revisions to the code before the conference. After I make my edits to the paper I might post it on this site. It's not very long, only 5 pages, but it's still pretty technical. Still, some people might be interested in reading it just because the idea of computers composing music sounds a little strange and it's pretty cool to see actual progress being made in that field.
This is my first time presenting at an academic conference, so if anybody has any experience with this and has some pointers or advice I would be very interested.