Image Processing and Neural Networks Lab

I. Goals of Theoretical Work

A. Bounds on training error for neural networks,

B. Algorithms for calculation of these bounds,

C. New neural network structures (paradigms), and

D. Improved algorithms for neural network sizing (complexity estimation), training, feature selection, and pruning.

E. Large Signal/Image Processing Systems Using Part D Algorithms as Building Blocks

F. Products

1. technical papers

2. theses

3. dissertations.

II. Applied Research

A. Summary

The main product of our applied research has been a series of software packages based upon the theoretical work described above. Examples include generic software packages for developing;

1. Neural or conventional classification networks,

2. Neural or conventional mapping or estimation networks, and

3. Bounds on neural network training error.

B. Application-Specific Software Systems Developed

1. Automatic target recognition,

2. Electro-telluric wave signal processing,

3. Estimating exponentials for the pulsed neutron porosity (PNP) well-logging tool.

4. Endpoint detection (for the semiconductor industry), and

5. Medium term load forecasting (for the electric power industry),

6. Very short term load forecasting (also for the electric power industry)

Click here to email any questions or comments about this web site.

Copyright © 2004 Image Processing and Neural Networks Lab
Last modified: March 06, 2007