What does this research mean for the field?

The BiLSTM-LSTM neural network achieves a mean square error of 0.0001 and a coefficient of determination of 0.9972 in modeling the S-parameters of GaAs pHEMT. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

This research aims to enhance the accuracy of modeling S-parameters for GaAs pHEMT using BiLSTM-LSTM neural networks.

March 10, 2026

Inverse Modeling of S‐Parameters for GaAs pHEMT Based on BiLSTM ‐ LSTM

Key Points

This research aims to enhance the accuracy of modeling S-parameters for GaAs pHEMT using BiLSTM-LSTM neural networks.
Utilized BiLSTM-LSTM neural network for modeling
Focused on inverse modeling of S-parameters
Analyzed the relationship between parameters and their predictions
Evaluated performance using mean square error and coefficient of determination
Achieved an optimal mean square error of 0.0001
Reported a coefficient of determination (R²) of 0.9972
Found an optimal mean absolute error of 0.0049
Confirmed superior performance in modeling nonlinear relationships

Abstract

ABSTRACT To further enhance the precision of transistor modeling, the bidirectional long short term memory‐long short term memory (BiLSTM‐LSTM) neural network is utilized for the inverse modeling of the scattering parameters (S‐parameters) for the gallium arsenide pseudomorphic high electron mobility transistor (GaAs pHEMT) here. The modeling results show that the optimal mean square error (MSE) is up to 0.0001, the coefficient of determination ( R 2 ) is 0.9972 with the optimal mean absolute error (MAE) of 0.0049. Therefore, BiLSTM‐LSTM has superior advantage for the nonlinear relationship modeling for the S‐parameters of GaAs pHEMT.

Bookmark

Inverse Modeling of S‐Parameters for GaAs pHEMT Based on BiLSTM ‐ LSTM

Key Points

Abstract

Cite This Study