Research

Well, it’s been a long time since I’ve done actual research. I still have hopes of starting something up again … maybe digging deeper into my failed attempt at extracting knowledge from neural nets or something around logical expressions and understanding their complexity. Who knows. At this point all I have is some, what seems like, ancient lore.

Thesis

Knowledge Extraction from Neural Nets …

My masters thesis was entitled Knowledge Extraction from Neural Nets through Graph Theoretical Algorithms. This research was somewhat related to my previous efforts into Tournament Isomorphism. It took a while, but after being drafted, re-drafted, edited, re-edited, formatted and re-formatted it is now available online.

Acknowledgments

Thanks to my Thesis committee: Marty Wolf, Jon Hakkila and Richard Roiger. Also thanks to David Haglin for also lending an ear and being willing to serve on the Thesis committee.

Tournaments

Tournament Isomorphism

Hi, so you are interested in tournaments or you’d like to more about them. As an undergraduate I was fortunate enough to be selected to help out with some research conducted here at MSU. I then became very interested in the topic and decided to spend some more time on it. Anyway, I ended up writing a paper with the title Convex subsets as a heuristic in tournament isomorphism. As part of the research I also gave a poster presentation of this research at the SCCS or Small College Computing Symposium in April of 1996 at St. Cloud State University in Minnesota (stale link to conference website).

Acknowledgments

This research is based on an algorithm found by Dr. Marty Wolf and Dr. David Haglin and was funded by the computer science department at Mankato State University through a grant from Hickory Tech … THANK YOU

Database Mining

As my first research topic as a graduate student I explored Database Mining with my graduate advisor Dr. Richard Roiger. Here is a list of some links related to this topic.

The paper I wrote as part of this research is available in html and postscript. To be a little more specific, the paper deals more with explanation facilities in neural networks, but I was interested in how to apply neural networks in a useful way to database mining. Since there are a lot of people (me included) who have a hard time in taking output from a black box (the neural net) I figured it may make for interesting exploration. As a matter of fact I’m right now very interested in the application of convex subsets in p-partite tournaments, but more of that latter.

Other Links

There is a database of papers, which are available online and Neuro Net has lots of info on neural nets (and since the site is in Europe, hence the name … clever, eh).