Kim Houck

Department of Computer Sciences

Possible topic talk and project ideas for CS394N (Neural Networks)

My main research area is machine learning. I also have a background in engineering and robotics, including some computer vision and control systems experience. My main interest is in learning for robotic sensing and control. Additional topics that I find interesting include the modeling of natural systems, cognitive science and developmental learning, to name a few. I also worked with Support Vector Machines for my masters thesis, although I would like to broaden my horizons at this point.

Some topics that I'd potentially be interested in for the topic talk are (in no particular order):

(a) Applications of neural networks (for robot control, low level sensory processing or other applications)

(b) Hebbian learning and self-organization.

(c) Learning algorithms that grow and delete units.

(d) Methods for evolving neural network topologies.

Most of my programming experience has been in C/C++ on Linux, but I also have experience with other langauges (i.e. Python, Java, Matlab, etc) and could work with additional languages if nessesary.

I am working with Sean Trettel.