.png){kind=icon}
.png){kind=icon}
Return
Neuro-Impedance Function for Electromagnetic Modeling
.png){kind=icon}
PRE
AI
DOI：10.1109/MMM.2024.3486588.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Artificial neural networks (ANNs) can predict electromagnetic (EM) responses under different parameters quickly and accurately while avoiding time-consuming simulations. Knowledge-based neural network (KBNN) techniques incorporating previous knowledge improves model learning and speeds up development. When accessing such knowledge is difficult, the KBNN can employ neurofunction approaches. The existing neurofunction methods tend to deal with the issues of order changing and parameter discontinuity simply from a mathematical perspective. This article is the latest addition to the review of neurofunction methods, which can be summarized as neuro-impedance function (neuro-IF) EM modeling technology. Unlike previous methods, the new neuro-IF solves the problem essentially by changing the physical meaning of the function, which fundamentally avoids the issues of order changing and parameter discontinuity. This new idea further expands the modeling capability boundaries of neurofunction methods and provide a new inspiration for future research.
Keywords:
Training
Reviews
Optimization methods
Predictive models
Feature extraction
Mathematical models
Vectors
Parametric statistics
Propagation constant
Usability
Journal
IF:
2.6
Papers:
3.2K
Citations:
3.1K