Investigating the Fault Tolerance of Neural Networks
Neural Computation
Artificial neural networks: a review of commercial hardware
Engineering Applications of Artificial Intelligence
Tolerance of radial basis functions against stuck-at-faults
ICANN'05 Proceedings of the 15th international conference on Artificial neural networks: formal models and their applications - Volume Part II
Complete and partial fault tolerance of feedforward neural nets
IEEE Transactions on Neural Networks
Hi-index | 0.00 |
This paper presents a technique for improving the fault tolerance capability of Artificial Neural Networks. This characteristic of distributed systems, which is usually pointed out as one of the advantages of this structure hasn't been deeply studied and can be improved in most of the networks. The solution implemented here consists of changing the architecture of feedforward artificial neural networks after the training stage while maintaining its output unchanged. It involves evaluating the elements of the Artificial Neural Network which are more sensible to a fault and duplicating inputs, bias, weights or neurons, according to the evaluation done before. This solution is very interesting because it allows maintaining the pre-trained network, but its cost is the need of additional hardware resources to implement the same network. The paper also presents an example of the application of the technique to illustrate its effectiveness.