Recurrent neural networks can learn simple, approximate regular languages.

Mikel L. Forcada, Antonio M. Corb-Bellot, Marco Gori, Marco Maggini

Browse the full IJCNN paper archive.