An accelerated learning algorithm for multilayer perceptron networks
- Texas A M Univ., College Station, TX (United States). Dept. of Nuclear Engineering
- Univ. of Texas, Austin, TX (United States). Dept. of Mechanical Engineering
- Cairo Univ. (Egypt). Dept. of Electrical Engineering
- Univ. of California, Irvine, CA (United States). Dept. of Electrical and Computer Engineering
An accelerated learning algorithm (ABP--adaptive back propagation) is proposed for the supervised training of multilayer perceptron networks. The learning algorithm is inspired from the principle of forced dynamics'' for the total error functional. The algorithm updates the weights in the direction of steepest descent, but with a learning rate a specific function of the error and of the error gradient norm. This specific form of this function is chosen such as to accelerate convergence. Furthermore, ABP introduces no additional tuning'' parameters found in variants of the backpropagation algorithm. Simulation results indicate a superior convergence speed for analog problems only, as compared to other competing methods, as well as reduced sensitivity to algorithm step size parameter variations.
- DOE Contract Number:
- FG02-89ER12893
- OSTI ID:
- 6855104
- Journal Information:
- IEEE Transactions on Neural Networks (Institute of Electrical and Electronics Engineers); (United States), Journal Name: IEEE Transactions on Neural Networks (Institute of Electrical and Electronics Engineers); (United States) Vol. 5:3; ISSN 1045-9227; ISSN ITNNEP
- Country of Publication:
- United States
- Language:
- English
Similar Records
Early detection of incipient faults in power plants using accelerated neural network learning
Improving learning of neural networks for nuclear power plant transient classification