Bioinformatics.fr



<< Back

Suggest a laboratory for FREE

Joe Song Lab

Location :Department of Computer Science at the New Mexico State University, Las Cruces, New Mexico, USA.

http://www.cs.nmsu.edu/~joemsong/

PageRank  out of 10

Activities:

My research interest is concerned broadly with the interface between computer science and statistics. Based on principles of summarizing observations, statistics determines a quantity to be computed from observed data and evaluates how significantly the quantity is supported by the data. Computer science pursues efficient algorithms to compute the quantity. Challenging problems often involve both. I develop efficient statistical modeling and learning algorithms to compute effectively a mathematical representation of the underlying mechanism that generates the observed data. I aim at mathematical representations that delineate the underlying mechanism by the structural, temporal, and statistical dependencies among the many variables in the observations.

I am intrigued by computing applications in life sciences due to personal experiences and recent quantitative research trends in life sciences. The attachment started when I first participated in a research project on creating a single photon emission computer tomography machine as an undergraduate student. Since then, I have designed a variety of algorithms and applied them in bioenergy, cancer research, microbial ecology and neuroscience.

The following is a list of my ongoing research initiatives:

  • Statistical computing--My core research is to determine the statistical, functional, and temporal dependencies among from hundreds to thousands of random variables.
  • Computational systems biology--I m working on modeling gene regulatory networks using discrete dynamic systems and generalized logical networks.
  • Neuronal signal analysis--I have developed data stream clustering algorithms to analyze spike signals emitted from neurons.
  • Computer vision--I believe in approaches that solve computer vision problems with integrated low-, mid-, and high-levels are the most promising.


Cancel