A simple learning of binary neural networks with virtual teacher signals

M. Shimada, T. Saito
IJCNN'01. International Joint Conference on Neural Networks. Proceedings (Cat. No.01CH37222)  
This paper presents an efficient geometrical learning algorithm for binary neural networks. Our supervised learning algorithm includes flexible and speedy linear separation method using virtual teacher signals. As compared with conventional algorithms, our algorithm can reduce the number of hidden layers and improve the variation of the connection parameters for complex teacher signals. The learned networks are suited for hardware implementation
doi:10.1109/ijcnn.2001.938480 fatcat:nlgcjumx7fcsdcub6eef5qvthu