Description
We present a novel training algorithm for a feed forward neural network with a single hidden layer of nodes (i.e., two layers of connection weights). Our algorithm is capable of training networks for hard problems, such as the classic two-spirals problem. The weights in the first layer are determined using a quasirandom number generator. These weights are frozen---they are never modified during the training process. The second layer of weights is trained as a simple linear discriminator using methods such as the pseudo-inverse, with possible iterations. We also study the problem of reducing the hidden layer: pruning low-weight nodes and a genetic algorithm search for good subsets.
Date of creation, presentation, or exhibit
5-26-1995
Document Type
Conference Paper
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Recommended Citation
Anderson, Peter G.; Gaborski, Roger; Ge, Ming; Lung, Mei-Ling; and Raghavendra, Sanjay, "Using quasirandom numbers in neural networks" (1995). Accessed from
https://repository.rit.edu/other/223
Campus
RIT – Main Campus
Comments
Proceedings of the International ICSC Symposium on Fuzzy Logic (1995) A50-A56 "Using Quasirandom Numbers in Neural Networks," Proceedings of the International ICSC Symposium on Fuzzy Logic. ICSC Academic Press. Held at the Swiss Federal Institute of Technology (ETH): Zurich, Switzerland,: May 26-27, 1995. The authors wish to thank Alex Mirzaoff and Eastman Kodak Company for support of this project. ISBN: 390-64-5400-2Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works in February 2014.