Multi-layer Perceptron based Comparative Analysis between CNTFET and Quantum Wire FET for Optimum Design Performance

Abstract

A novel classification technique is applied for identifying carbon nanotube FET and quantum wire FET based on their electrical characteristics and percentage error is estimated using multi-layer perceptron analysis to justify the accuracy of computation. Two different cross-validation methods, namely decision table and multilayer perceptron (MLP) are applied on same data set of both the devices, and results speak about higher accuracy when MLP is performed. Also, for different testing-training set of data, MLP performs far better than conventional decision table approach; when correlation coefficient, mean absolute error, root mean squared error, relative absolute error and root relative squared error are computed. For comparative study, similar geometrical configuration, and equivalent biasing arrangement of both the devices are assumed, and identical number of iterations is performed for equal subsets. Results speak supremacy of MLP technique applied for classification and identification of nanometric devices based on their electronic attributes.