Practical neural network recipes in C++
Practical neural network recipes in C++
Hi-index | 0.00 |
The problem of definition of parameters of thin anisotropic films used in a microelectronics on the basis of ellipsometrical measuring is explored. The method of definition of parameters of films with use of neuron networks is offered. The networks is trained in space of acceptable values of parameters of layered system. The algorithm of tutoring of a network grounded on a rule Widrow-Hoff. At tutoring the error of experimental data's was taken into account. The neuron networks is applied for definition of parameters of uniaxial films of Langmuir-Blodgett dimethy1-3,4:9,10-perylene-bis(dicarboximide). The network has shown high performance, the results coincide with obtained other methods. The network can be applied for examination of layered systems.