Advances in artificial neural network models of active devices

Abstract

This paper reviews some recent advances in the application of artificial neural networks (ANNs) to measurementbased modeling of active devices. For transistor models, the advent of the adjoint training method for terminal charges, and the training of constitutive relations depending on multiple dynamical variables -some identified from measured waveform data from nonlinear measurements -are surveyed. The ability to implement exact discrete symmetry constraints in ANN-based models is another example. Several examples of practical models implemented in commercial simulation tools are cited to demonstrate that ANN technology has become a mainstream tool for advanced measurement-based modeling of active devices. Areas for future development are also outlined.

Modeling of inversion layer capacitance of III-V double gate MOSFETs using a neural network-based regression technique

“…A machine learning-based predictive model for current-voltage and capacitance-voltage characteristics of FinFET is reported in [18]. Some other similar works related to device modeling using deep learning technique is reported in [19][20][21][22][23].…”

Research on Device Modeling Technique Based on MLP Neural Network for Model Parameter Extraction

“…As the complexity of the device model increases, the number of model parameters goes up. Manually specifying new model parameters for complex device physics equations is much harder than adding new variables to the NN model [9].…”