B. GOVOREANU, J. SUYKENS, W. SCHOENMAKER,
G. DIMA, J. VANDEWALLE,
M. PROFIRESCU
On the Use of Bayesian Learning Neural Networks for TCAD Empirical Modeling
Abstract.
In this paper we briefly discuss some of the existing response surface modeling
techniques currently used for TCAD purposes and bring to the readers' attention
a Bayesian framework for training feed-forward neural networks. We assess the
model performances of different models by studying a 0.25 μm nMOS transistor. We
show that the Bayesian learning neural network modeling may successfully replace
traditional models in particular cases. |