|
Jay Powell
Computer Science Graduate Student |
|
|
|---|---|---|
LinksHomePublications Teaching Miscellaneous Resume |
I am a PhD student in the Computer Science department at IU. I began my graduate career in 2005. My interests lie in the areas of case-based reasoning, knowledge based computation, machine learning, and data mining. I recently completed a Master of Science at IU, while my Bachelor of Science degree was completed at the University of Nebraska at Kearney. I have also spent several years designing and implementing software for Ward Laboratories, Inc. As an undergraduate I worked on a case-based reasoning (CBR) under the guidance of John Hastings. A significant problem for CBR systems involves acquiring domain knowledge that a reasoner can use to solve problems. This is commonly referred to as the knowledge acquisition bottleneck. Our initial worked focused on learning case bases from scratch using machine learning techniques. The learner would start off with no domain knowledge or cases (other than the ability to complete a few simple, random actions) and use reinforcement learning to guide the development of a case base. Later work was tested in the chess domain, where we added a basic domain-specific problem-solver to fall back on when the case base could not suggest an appropriate move to make. I am currently working with David Leake on a CBR project for my Ph.D. dissertation. Our work also focuses on the knowledge acquisition problem, where domain-relevant knowledge is discovered when the need arises by mining the Web. This work merges the knowledge discovery process with the classic case-based reasoning cycle. The acquired knowledge is used for adaptation (modifying cases). The goal of this project is to develop a generic case adaptation framework that is capable of reasoning in multiple domains with minimal pre-coded domain knowledge. |
|