Michael DeMan's Research Page
Graduate Student
Dept. of Computer Science
michael@deman.com
Note: These pages are circa 1996!
A newer link is Gemini Solutions.
Research Assistant
Currently I am funded by NOAA and
USAF for a research assistantship with the Machine Learning Group. here at W.W.U.
The Machine Learning Group is a loose association of Computer Science faculty and graduate students who are interested in problems related to machine learning and, in particular, the application of ML to problematic data analysis and interpretation.
The group works closely with members of the Institute for Watershed Studies and the Institute for Environmental Toxicology and Chemistry in analyzing data obtained from numerous ecological and toxicological studi
es.
Current Projects
My work here is in the area of data conversion, access and transparency. We have adopted
NetCDF
from UniData as our primary data format.
I am developing
NetCDF
If you're interested in any of these tools I have some of t
hem at my ftp site. Or send me some e-mail about it.
This past spring we have been developing the WWU Distance Learning Center - a project that will allow students on any platform that supports a JAVA enabled web browser to take classes here at WWU via the internet.
Our prototype project has been focused towards allowing students at Olympia Community College to take computer science 221 via their PC lab. The prototype allows the instructor to broadcast to students via various chat functions as well as answer questions individual and/or groups of students may have.
The students learn to program via a DLC controlled telnet session into our Sun lab here at WWU. The instructor can view various student's screens during the telnet session as well as take control of their session to help them or broadcast a telnet session of his own to everyone to illustrate an example.
We intend to have the prototype functioning by the end of summer and available to Olympia Community College for the fall quarter.
WWW and Object Oriented Programming
This spring was trying to get up to speed with JAVA, distributed objects and interactive programming on the Web. I've put together a few review pages for
OpenDoc,
JAVA,
Web Obects and CORBA,
COM/OLE and the
three tier client server model.
This thesis will explore the applicability of Markov models to non-discrete multivariate data sets. Recent work in natural language processing [1] and speech processing [2] have shown traditional and other types of Markov models to be extermely powerful tools. In statistical language learning each letter ('a' through 'z' plus punctuation) is considered to be a single symbol. Using various techniques the model 'learns' from a base set of text and generates probabilities for determining what the next symbol should be given the symbols previously seen. Typically the Markov model looks at the current state as well as the past two or three states when generating its probabilities. How far the model 'looks back' is called the memory length of the model.
Star Game Object Model
TL1150/OSL1150
- Development and continuing support of software to provide real-time data acquisition and systems control for the TL1100/1150 and OSL1150 themolumniscent geological dating equipment.
tl1150 patches
Related Sites
- CSLI / Stanford University
- The Hebrew University Machine Learning Lab
Thanks for the visit!
personal | thesis | D.L.C.
