Title :
Algorithm for pruning hidden units in multilayered neural network for binary pattern classification problem
Author :
Watanabe, Eiji ; Shimizu, Hikaru
Author_Institution :
Dept. of Inf. Process. Eng., Fukuyama Univ., Japan
Abstract :
This paper proposes a simple algorithm for pruning hidden units and introduces a cost function which can detect the unnecessary hidden units for learning patterns and prune them from the hidden layer. From simulation results, it is shown that the proposed algorithm can easily prune hidden units.
Keywords :
feedforward neural nets; learning (artificial intelligence); optimisation; pattern classification; binary pattern classification; cost function; hidden units pruning; learning patterns; multilayered neural network; Convergence; Cost function; Information processing; Intelligent networks; Monitoring; Neural networks; Pattern classification;
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.1993.713923
