Glen Upton's Senior Project Journal

Sept 9th: Researching for something in Psychology or Music, still looking for topics.

Sept 12th: Centered on learning algorithms, more research, preparing for paper.

Sept 17th: Game Theory seems interesting for learning algorithms, more research. I wish I could find more things in print, instead of citations on the web.

Sept 25th: Gave Survey presentation in class, felt rushed, scattered, too much information to cover. Found interesting topic on Swarm Intelligence while waiting in the library for class. You always find the best things when you're not looking.

Sept 26th: Jim showed me a good resource for finding papers. Already printed one paper out.

Sept 28th: Revamped page to reflect change in topic to Swarm Intelligence, I think I'm going to stick with this one, there is plenty of information on it, and it's interesting.

Oct 1st: Stopped researching for a while so that I could write my survey report, which gives a general overview of swarm intelligence and its applications.

Oct 9th: Recently installed Linux Red Hat on my laptop to help facilatate the production on some swarm software on my own system, without running the risk of messing up the entirity of the ACL network. Plus, this way I can keep my work indefinitely.

Oct 10th: Received several books through interlibrary loan that will prove to be useful. Jim gave me a book full of NP complete problems to look through to see if any of them might be interesting to solve using swarm intelligence.

Oct 12th: Installed swarm software on my laptop, but have yet to get it working, this might be a problem.

Oct 16th: Closing in on a proposal for using an ant algorithm to solve an NP complete algorithm. I'm ready for mid-semester break.

Oct 19th & 20th: Spent mid-semester break in Maryland, had plenty of time in the car to think about a project. I think I want to create an ant colony algorithm to find an optimal solution to the travelling salesperson problem. I'd want to make a GUI too that would allow the user to change pheromone evaporation rates and number of ants and the weights on the edges of the graph.

Oct 22nd: I had sent Jim an e-mail telling him about my idea for the project. He thinks I should choose a different NP complete problem instead of the TSP. I like the TSP because it is familiar to everyone, and I won't have to waste time researching the problem and explaining it to people. For me, it doesn't matter what the problem is, because it is just an analysis of the algorithm. I never thought I would be interested in algorithmic analysis.

Oct 29th: Last week Jim had suggested that I work on an algorithm called the Stable Roommates problem. It's a matching problem which is just a subset of the Stable Marriage problem. In order to realize the Roommates problem as solveable by an algorithm, I need to think of a way to represent the Roommates problem in terms of a graph. Ants work best with graphs, for then they can utilize their ability to find shortest distances. I'm having trouble finding information on the Roommates problem, especially when it comes to thinking of it as a graph. It's getting down to the wire since the proposal papers are due next week. We'll see what happens.

Nov 7th: Recently gave proposal paper presentation. It went well and the concept of my ant colony algorithm that solves the Stable Roommate problem is coming together. Jim and I went over things that need to be refined. Now I need to prove that my algorithm will work, this is probably the hardest step.

Nov 14th: Jim has helped me with the proof for my algorithm. The stables roommates problem is an NP complete problem. I need to tidy up my proof and begin coding this thing. I'm also going to be working on a GUI for the algorithm as well.

Nov 21st: Although I'm on break, I have began coding up the problem, and I've gotten quite a lot finished. The structure of the problem is finished. I'm using an adjacency matrix to describe the graph.

Dec 2nd: Back at school and not much time left, I have to get my butt in gear.

Dec 9th: I've finished coding up the algorithm, it has a few bugs that needs to be worked out.

Dec 12th: After watching half of the class present, I started working on my GUI for the algorithm, which is finished and working. The gui is going along smoothly, I've had to sacrifice somethings because of lack of time, but I like where its going.

Dec 15th: I presented today, and I had to go first. I thought it went very well. After the presentation it became apparent to me that my algorithm had not quite solved the problem, but I am still proud of the work I have done this semester.