Nonlinear Regression of characteristic of gunn diode: A Neurocomputing Approach

Mrs. S. N. Kale, Dr. S.V. Dudul


Non-linearity is observed in the transfer characteristics of gunn diode. Neural Network can elegantly solve a typical regression of characteristic of gunn diode. The dataset is obtained by performing experiment on a typical continuous wave gunn diode MA49156-30. The numbers of readings are regarded as samples. The dataset is obtained, which is used for regression. After rigorous computer simulations authors develop an optimal MLP NN model, which elegantly performs such a nonlinear regression. Results show that the proposed optimal MLP NN model has a MSE as low as 4.24 x 10-5,correlation coefficient as high as 0.9852 when it is validated on the cross validation dataset.


Keywords: Regression, MLP NN, RBF NN, Jordan Elman network, Statistical model.

Full Text:




  • There are currently no refbacks.

Copyright (c) 2016 International Journal of Advanced Research in Computer Science