Fall 2003 Research Positions
Three positions are available for students to work with faculty on research projects. These are great opportunities to explore a research topic in depth with the guidance of a professor. Sophomores and beyond are welcome to apply; a work/study award is not necessary for this position. Positions range from 5 to 10 hours a week, at $7.05 an hour. Details for the three positions are detailed below.
Applications are due Friday, September 5. To apply, complete the CSB/SJU Student Employment Application Form . In the ``Position Applying for'' blank, write ``research assistant for X'' where X is the names of the professors guiding the projects in which you are interested. (You can apply for several that interest you at once.) Also, submit a paper résumé to Prof. John Miller, PEngl 213.
If you have questions about the application process, contact Prof. Miller ([email protected], PEngl 213, x3155).
Dr. Carl Burch: Learning in continuous environments
Imagine that you want to learn how to win a game consistently, but you can't really tell whether you're winning until you've won. Or imagine that you're a dog, and your owner is clearly trying to train you do to something, but you don't know what it is until you get your treat. Such learning situations, where the feedback to actions is delayed, are difficult but important. Computer scientists have thought of several algorithms for learning in these situations. One algorithm, temporal difference (TD) learning, has been particularly successful and has even been used to train the world champion backgammon player (a computer program).
In this project, we'll explore ways of applying TD learning to continuous environments. TD learning works by assigning values to different situations the learner can reach, refining them as the learner continues. Continuous environments present challenges because they present an infinite number of situations. Extending the work of a 2003 Honors Thesis (by Chris Marsh), we'll work on ways to represent these values and to refine the learner's understanding of them over time.
This project is for students talented in both mathematics and computer science. You will learn about current machine learning research, both through discussion with me and through library research. You will develop a program to test our concepts. And you will write and present a paper at a research conference. You can work from 6 to 10 hours a week; CSCI 200 (or current enrollment) is a prerequisite.
Dr. Chris Lusena
Frequently one has many tasks to do with many variations. One might consider how long the tasks will take or when they should be done by. One way of choosing which task to do is to do the most important one next. The structure used to do this is called a priority queue. Priority queues turn up in numerous computer science contexts so efficiency priority important. The purpose of this project is to investigate a new, potentially better way of implementing a priority queue. Research on this project will consist of library research on the data structure in question, implementing the new implementation and known implementations, and then running some experiments to compare the various implementations.
Dr. Noreen Herzfeld: The Social Implications of Computing
Noreen Herzfeld is looking for a research assistant to do bibliographic and library research in the area of social and theological implications of computing. The position would be for 5 hours per week and would consist primarily of locating and reviewing articles on topics such as the social implications of AI, the internet, computers and moral agency, and computers and ecological responsibility.