Matt Kretchmar
After graduating from Penn State with a degree in Computer Engineering and a minor in Philosophy, Dr. Kretchmar worked as a software engineer at IBM to develop their first data warehousing project. In his graduate programs at Rensselaer and Colorado State, Dr. Kretchmar focused on a variety of artificial intelligence and machine learning techniques. His Ph.D. dissertation analyzed a robust (fault tolerant) reinforcement learning controller for a large HVAC system. Dr. Kretchmar teaches a wide range of courses across the computer science curriculum as well as introductory liberal arts mathematics courses. Dr. Kretchmar’s classes often experiment with non-traditional pedagogies including a portfolio based system in his Sophomore Data Structures class, and a research paper based Artificial Intelligence seminar. He is also very interested in writing pedagogy and in first year student experiences; he served as Denison’s Dean of First Year Students from 2007 to 2012.
Research
My research area is machine learning techniques. I concentrate in Reinforcement Learning, especially in building controllers for various dynamic systems. Additionally I work in the area of classification techniques including Kernel Machines and Support Vector Machines. I also dabble in games and game theory, and in discrete and combinatorial mathematics.
Works
Selected Publications:
- Suspense at the Ballot Box. (with Nat Kell) The College Mathematics Journal, Vol 44, No 1. 2013.
- Tree Traversals and Permutations. (with Todd Feil and Kevin Hutson) Congressus Numerantium, Vol 172. 2005.
- Improved Automatic Discovery of Subgoals for Options in Hierarchical Reinforcement Learning. (with Todd Feil and Rohit Bansal) Journal of Computer Science and Technology. October, 2003.
- A Neighborhood Search Technique for the Freeze Tag Problem. (with Dan Bucatanschi, Blaine Hoffman and Kevin Hutson) Extending the Gap: Advances in Computing, Optimization, and Decision Technologies. 2007.
Other
Selected Student Research Projects:
- Text Message Authorship Classification Using Support Vector Machines, Yifu Zhou, 2013.
- A Reinforcement Learning Robotic Arm Controller, Taylor Kessler Faulkner, 2013.
- An Analysis of Ballot Ordering for Final Tribal Councils in the Television Series Survivor, Nat Kell. 2010.
- Kernel Methods for Image Processing, Dan Bucatanschi, 2006.
