Sign in
Neural network generalization using error correcting codes
Journal article   Peer reviewed

Neural network generalization using error correcting codes

Khalid Al-Mashouq
Digest of technical papers - IEEE International Conference on Consumer Electronics, pp.118-119
01/01/1994

Abstract

Robustness of a neural net relies on generalization which is its response to novel data that has not been trained with. Here we propose the use of error correcting codes to encode the output of a neural net in order to improve generalization. Simulation examples with the problem of speech recognition demonstrate that better generalization can be achieved using error correcting codes.

Metrics

4 Record Views

Details