Since 1993, Devices based on CNTs have applications
ranging from nanoelectronics to optoelectronics. The
challenging issue in designing these devices is that the
nonequilibrium Green's function (NEGF) method has to
be employed to solve the Schrödinger and Poisson
equations, which is complex and time consuming. In the
present study, a novel smart and optimal algorithm is
presented for fast and accurate modeling of CNT fieldeffect
transistors (CNTFETs) based on an artificial neural
network. A new and efficient way is presented for
incrementally constructing radial basis function (RBF)
networks with optimized neuron radii to obtain the
estimator network. An incremental extreme learning
machine (I-ELM) algorithm is used to train the RBF
network. To ensure the optimal radii for incremental
neurons, this algorithm utilizes a modified version of an
optimization algorithm known as the Nelder-Mead
simplex algorithm. Results confirm that the proposed
approach reduces the network size for faster error
convergence while preserving the estimation accuracy. Manuscript profile