Improvement Techniques for the EM-Based Neural Network Approach in RF Components Modeling

  • Authors:

  • Liu Tao;Zhang Wenjun;Ma Jun;Yu Zhiping

  • Affiliations:

  • Institute of Microelectronics, Tsinghua University,;Institute of Microelectronics, Tsinghua University,;Dept. of Computer Science, Xi'an Univ. of the Finance and Economics, Xi'an 710061, China;Institute of Microelectronics, Tsinghua University,

  • Venue:
  • ISNN '07 Proceedings of the 4th international symposium on Neural Networks: Advances in Neural Networks
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Electromagnetic (EM)---based neural network (NN) approaches have recently gained recognition as unconventional and useful methods for radio frequency (RF) components modeling. In this paper, several improvement techniques including a new data preprocessing technique and an improved training algorithm are presented. Comprehensive cases are compared in this paper. The experimental results indicate that with these techniques, the modified model has better performance.