Penacee: A neural network system for recognizing on-line handwriting.

I. Guyon, J. Bromley, N. Matic , M. Schenkel, and H. Weissman.
In E. Domany, J. L. van Hemmen, and K. Schulten, editors, Models of Neural Networks, volume 3, pages 255--279, Springer.

We report on progress in handwriting recognition and signature verification. Our system, which uses pen-trajectory information, is suitable for use in pen-based computers. It has a multi-modular architecture whose central trainable module is a Time Delay Neural Network. Results comparing our system and a commercial recognizer are presented. Our best recognizer make three times less errors on hand-printed word recognition than the commercial one.

Keywords: handwriting recognition, neural networks, on-line handwriting recognition, signature verification, Time Delay Neural Network, hybrid systems, pen-based computers, cursive handwriting.

[ next paper ]