One of the goals of artificial intelligence is to be able to view the computer as a “black box”, you simply give it the problem you want to solve, and it gives you the answer, without you needed to understand all of the internal workings. Evolutionary computation seeks to create this black box by harnessing the power of Darwinian evolution to solve computational problems.  Instead of programming a solution, the user simply initializes a population of very simple (and probably very bad) solutions, and then sits back while the population evolves until a good solution appears.  However in order to find a good solution, the parameters and dynamics of the evolutionary system have to be set appropriately, and to set them appropriately we have to understand the dynamics in the first place. Therefore my research has two sides, the first is to harness the power of evolution to solve real problems, and the second is to use computational power to study and understand the dynamics of evolution itself. This summer I will be continuing work on several different projects and students will be able to choose which interests them most. The three main possibilities are:

1. Diversity/Ecosystems - looking into how diversity evolves, how a population maintains it, how speciation occurs and how organisms come to specialize in specific tasks, and especially how diversity in an Evolutionary Algorithm's population can help it evolve better solutions more quickly.

2. Altruism - exploring how altruism evolves and how it is maintained in a population.

3. Evolving cooperative behavior in actual physical robots using neural networks.

Students interested in any (or all) of these areas are encouraged to apply for summer research.  Students should have taken Data Structures, and background in biological evolution is helpful but not required.  Classes such as AI or EC/A-Life would also be helpful background, but not necessary.

Project:  Online Collaborative Communities: Enriching and Understanding the Entry of Cited Information into Wikipedia (Dave Musicant, 2-4 students)

Wikipedia is amazing, isn't it?  Lots of people have come together to discuss content and produce a free encyclopedia whose articles are often among the top hits at Google. Wikipedia is often incomplete or incorrect, however, and the need remains strong to get accurate and current information into it. With a relatively mature body such as Wikipedia, this can be challenging to do, and new Wikipedia editors often feel daunted and leave the community. For this project, we'll look at developing tools whose purpose is to help get more content into Wikipedia, and to help Wikipedia editors stick around and be more productive. We'll also be looking at the nature of citations in Wikipedia, and how certain sorts of information are more effective than others for introducing content into Wikipedia. The project will be a mix of data analysis, recommender-system techniques, user interface design, and effects studies; we'll actually be looking to build tools, and measure the effects of using those tools on the people who use them. This project is part computer science and part social science; we'll be using computing algorithms and techniques to make an impact in how people work and interact.  Students interested in this project should have completed CS 201 (Data Structures).  No other courses are required, but it would be helpful to have taken one or more of CS 321 (Artificial Intellegence), CS 324 (Data Mining), or CS 334 (Databases).