Artificial neural networks in the solution of inverse electromagnetic field problems

The use of artificial neural networks in the solution of inverse electromagnetic field problems is investigated. It is shown that artificial neural networks, while being no panacea, have a role to play in a limited domain of applications-that is, while it is ineffective to train networks to cover a broad class of devices, it is indeed possible to develop well-trained networks that function effectively over a narrow range of performance of a particular class of device. Particularly if one knows the desired geometry approximately and uses training sets around this geometry, simple neural networks with a few training sets can be used to do an effective job. However, neural networks cannot be used efficiently without such prior knowledge.>