Title :
Training strategies for weightless neural networks
Author :
Ludermir, Teresa B. ; de Olivereira, W.R.
Author_Institution :
Dept. de Inf., Univ. Federal de Pernambuco, Recife, Brazil
Abstract :
Weightless neural networks (WNN) are implemented as random access memories. Training WNN requires only global error signals. WNN simulations can learn significantly faster than learning by error-backpropagation. The aim of this paper is to discuss different training strategies for WNN. One new strategy is suggested.
Keywords :
learning (artificial intelligence); neural nets; probabilistic automata; random-access storage; cut point node; global error signals; learning; probabilistic automata; random access memories; weightless neural networks; Acoustic propagation; Artificial neural networks; Character recognition; Computational modeling; Computer networks; Formal languages; Neural networks; Neurons; Random access memory; Read-write memory;
Conference_Titel :
Neural Networks, 1993. IJCNN '93-Nagoya. Proceedings of 1993 International Joint Conference on
Print_ISBN :
0-7803-1421-2
DOI :
10.1109/IJCNN.1993.714288
