Review of neural network technique for modeling PA memory effect

In this article, we provide an overview of the applications of ANN to efficient behavioral modeling of PA with short and long term memory effects. Summary of various PA memory effect modeling techniques including dynamic neural networks (DNN) and recurrent neural networks (RNN) are discussed. These techniques extract slow-changing signals from the inputs and outputs of the PA and use these signals as extra inputs of models in order to effectively represent long term memory effects. The models can be trained directly with only the input-output data without having to know the internal details of the PA.