Thomas E. Kammeyer



About me...

I was a graduate student at the University of California at San Diego in the Computer Science and Engineering Department. Then I graduated. Now I may be found at kemosabi {that's the user name) (type shift-2] reductio <that's the second-level domain) (now type a radix separator> com (that's the TLD). I only wrote that in last sentence in that way because it should make it nominally harder to harvest the address from this page. The mail harvesters are getting wise to the whole spell-the-symbol trick.

My research interests are in machine learning; my thesis was concerned with the use of evolutionary algorithms for grammar induction. For the past several years, however, I've worked in biotechnology on software for molecular biology lab instrumentation and on algorithms to interpret and analyze the output of such instrumentation and on cluster and desktop code for a medical device that delivers therapeutic radiation to treat cancer.


A copy of my thesis is available via anonymous FTP at ftp.cs.ucsd.edu as pub/guest/tkammeye/TEKthesis.ps. Note that this is uncompressed postscript, size ~1.5M. You can also just click on the link a couple of sentences back. Linked here and in the ftp directory, you will also find the following...

A version of my FOGA4 paper on EvoGrams, an evolutionary algorithm for stochastic context-free grammar in the ftp directory. Note that this is a version before edits were applied based on reviews, so the version in the FOGA4 book would be better to read. This version will suffice if you can't get the proceedings in a timely fashion. The algorithm description influenced chapter IV of the paper, and the results are folded into chapter V, notably the results on learning and evolution in section G.

A slightly longer, though less polished, version of an Evolutionary Programming '98 article on maintaining known good solutions in EAs. This material found its way into the thesis in chapter IV and chapter V, section F.

Not coming soon unless someone screams loud enough to get me to post it...
An article on evolving CMPX networks should be available here soon. Chapter VI of the thesis covers the same material. The journal article appeared in Artificial Life v2n2.

Finally, a FOGA3 article on developmental models in genetic algorithms that I coauthored with Bill Hart and Rik Belew is available. This article contains a lot of the basic ideas that motivated my thesis work and contains some early parts of the evolving sorting networks work in chapter VI of my thesis.

For anyone interested, the collection of ideas I wish I could've worked on had there been time is in chapter 8, section C of my thesis. The one that strikes me as most interesting is in VIII.C.1 and is entitled "Derivational EvoGrams".
I was also the caretaker of GAucsd, a Genetic Algorithm package maintained locally.

Interesting places to surf at UCSD...