Sandia Mountains

Dr. John Burge

Post-doc

The University of New Mexico
Department of Computer Science
Farris Engineering Bldg. 329
Albuquerque, NM 87131-1386
lawnguy at cs.unm.edu
505-277-8432

Sandia Mountains


I was born and raised in Albuquerque, New Mexico where I attend the University of New Mexico. I received my BS in Computer Science as well as my Masters of Computer Science from UNM. I recently completed my PhD in computer science under the advisement of Dr. Terran Lane.

RESEARCH

I focus my research in the wonderful world of machine learning.  Currently, my interests deal with Bayesian nets, and methods for automatically finding the “best” Bayesian network structure for a given set of data.  Applying standard search techniques to networks modeling functional magnetic resonance imaging (fMRI) data, I have found strongly differing relationships between the amygdala and much of the brain in patients afflicted with mild to moderate senile dementia.  Extending this preliminary work, I am working on methods for improving the discriminative abilities of Bayesian networks I have employed as well as methods to incorporate prior domain knowledge, in the form of a hierarchical arrangement of random variables, to improve both structure search and parameter learning.

·         Curriculum Vita

 

The link provides a web version of my CV. As MS Word version can be retrieved here.

 

·         Reading list

 

This list is somewhat outdated, this list deals primarily with relational learning papers.  While that is certainly a fascinating area, my doctoral studies eventually took me down a different path and I did not keep up with the relational learning field.

 

·         PhD Thesis

 

This is a link to my PhD thesis.  The zip file contains both the original Word 2007 format document and some converted pdf files.  Unfortunately, during the conversion to pdf, some of the equations were converted to bitmaps (why oh why would they do that?!), so if you have the ability to read the newer Word .docx format, the quality would be higher.  I hope to find a resolution to that problem soon.

 


 

SOFTWARE

 

SBNet

 

During the course of my PhD doctoral studies, I developed a software package that (among other things) learns the structure for dynamic Bayesian networks (it has functionality to work with normal Bayesian networks as well, but that functionality is not as rigorously tested).

 

 

fMRI Toolbox

 

I also wrote some tools, the fMRI Toolbox, that allows the viewing of fMRIImages and the processing of those images into text files that summarizes the average voxel activity in about 150 different regions of interest (ROIs) as defined by the Talairach database.

 


Academic Activities

Reading Group: I attend a machine learning reading group which meets Wed, 3:00-4:30 during the summer 2007 semester. For more info, click here.

Wiki Link for the wiki Terran keeps on our Neuroimaging group (not much publically-accessible information, mostly for intra-group usage)

 


The University of New Mexico

Last updated July, 2007.