Deep Neural Network Technique for High-Dimensional Microwave Modeling and Applications to Parameter Extraction of Microwave Filters

Jin, Jing; Zhang, Chao; Feng, Feng; Na, Weicong; Ma, Jan; Zhang, Qi‐Jun

doi:10.1109/tmtt.2019.2932738

Cited by 163 publications

(112 citation statements)

References 40 publications

Supporting

Mentioning

112

Contrasting

Order By: Relevance

“…In order to reduce the number of hidden neurons as well as avoid the vanishing gradient problem, both rectified linear units (ReLUs) and sigmoid functions are utilized as activation functions in the hybrid deep neural network for high-dimensional microwave modeling. 42 Figure 5A shows the sigmoid function. It can be presented as: 43 σ γ where γ is the total input to the hidden neuron.…”

Section: Activation Functions For the Hybrid Deep Neural Network Tementioning

confidence: 99%

“…A hybrid deep neural network modeling method 42 is presented to address the high-dimensional modeling of microwave components. Figure 7 shows the structure of the hybrid deep neural network model, which is a fully connected neural network with many hidden layers.…”

Section: Structure Of the Hybrid Deep Neural Network Techniquementioning

confidence: 99%

“…41 Both techniques address the problem of nonuniqueness in inverse modeling. A recently developed hybrid deep neural network modeling technique 42 and the application to inverse modeling are reviewed in section 4. The hybrid deep neural network structure is presented.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Recent advances in neural network‐based inverse modeling techniques for microwave applications

Jin

Feng

et al. 2020

Int J Numerical Modelling

Self Cite

View full text Add to dashboard Cite

Inverse modeling of microwave components plays an important role in microwave design and diagnosis or tuning. Since the analytical function or formula of the inverse input-output relationship does not exist and is difficult to obtain, artificial neural network (ANN) becomes an efficient tool to develop inverse models for microwave components. This paper provides an overview of recent advances in neural network-based inverse modeling techniques for microwave applications. We review two different shallow neural network-based inverse modeling techniques, including the comprehensive neural network inverse modeling methodology and the multivalued neural network inverse modeling technique. Both techniques address the problem of nonuniqueness in inverse modeling. We also provide an overview of recently developed hybrid deep neural network modeling technique and the application to inverse modeling. For the inverse modeling problem with high-dimensional inputs, the relationship between the inputs and the outputs of the inverse model will become more complicated and the inverse modeling problem will become harder. The deep neural network becomes a practical choice. The hybrid deep neural network structure is presented. The recently proposed activation function, specifically for microwave application, and a three-stage deep learning algorithm for training the hybrid deep neural network are reviewed. K E Y W O R D S artificial neural network, deep learning, deep neural network, inverse modeling, microwave components

show abstract

Section: Activation Functions For the Hybrid Deep Neural Network Tementioning

confidence: 99%

Section: Structure Of the Hybrid Deep Neural Network Techniquementioning

confidence: 99%

See 1 more Smart Citation

Recent advances in neural network‐based inverse modeling techniques for microwave applications

Jin

Feng

et al. 2020

Int J Numerical Modelling

Self Cite

View full text Add to dashboard Cite

show abstract

“…A variety of ANN structures have been developed during past several years. Pure ANNs such as multilayer perceptron (MLP) neural networks [63], dynamic neural networks [64], [65], time-delay neural networks [66], recurrent neural networks [49], and the recently introduced deep neural networks [67] have been used for directly modeling the linear/nonlinear relationships between the model inputs and outputs.…”

Section: Artificial Neural Networkmentioning

confidence: 99%

Parametric Modeling and Optimization of Electromagnetic and Multiphysics Behaviors of Microwave Components

Zhang¹

View full text Add to dashboard Cite

Parametric modeling and optimization of electromagnetic (EM) and multiphysics behaviors are very essential parts of the design process of microwave components.For EM/multiphysics design, the computational cost of directly using EM/multiphysics simulator is very expensive because the simulation-driven design requires repetitive EM/multiphysics evaluations due to the adjustments of the values of design parameters. This thesis proposes new techniques to speed up the parametric modeling I would like to express my sincere appreciation to my supervisor, Prof. Qi-Jun Zhang for his professional guidance, constant support, invaluable inspiration, and suggestions throughout the research work and preparation of this thesis. His encouragement, motivation and thoughtful insights during my research made my journey in realizing my goal a memorable one. It was my honor to work under his supervision and guidance. I would also like to express my appreciation to my co-supervisor, Prof. Jianguo Ma for the encouragement and knowledgeable instruction. His professional suggestions provided a great help for me on my way pursuing my Ph.D degree. I would like to thank Tianjin University, where I have stayed for six years accomplishing my Ph.D studies. I would also like to thank Carleton University for the Cotutelle program for my Ph.D study.

show abstract

“…In [67], the space mapping (SM) concept has been elevated from pure EM parametric modeling to multiphysics parametric modeling. neural networks [68], dynamic neural networks (DNNs) [69], [70], radial basis function (RBF) neural networks [71], [72], recurrent neural networks (RNNs) [73], [74], [75], time-delay neural networks (TDNNs) [76], state-space dynamic neural networks (SSDNNs) [77], [78], and the recently introduced deep neural networks [79].…”

Section: Thesis Organizationmentioning

confidence: 99%

Advanced Parametric Modeling and Yield Optimization Approaches for Microwave Passive Components

Zhang¹

View full text Add to dashboard Cite

Parametric modeling and yield optimization techniques play important roles in electromagnetic (EM)-based microwave component design. In the first part of this thesis, we propose a novel training approach for parametric modeling of microwave passive components with respect to changes in geometrical parameters using matrix Padé via Lanczos (MPVL) and EM sensitivities. In the proposed approach, the EM responses of passive components versus frequency are represented by polezero-gain transfer functions. The relationships between the poles/zeros/gain in the transfer function and geometrical variables are learnt by neural networks. A novel

show abstract

Deep Neural Network Technique for High-Dimensional Microwave Modeling and Applications to Parameter Extraction of Microwave Filters

Cited by 163 publications

References 40 publications

Recent advances in neural network‐based inverse modeling techniques for microwave applications

Recent advances in neural network‐based inverse modeling techniques for microwave applications

Parametric Modeling and Optimization of Electromagnetic and Multiphysics Behaviors of Microwave Components

Advanced Parametric Modeling and Yield Optimization Approaches for Microwave Passive Components

Contact Info

Product

Resources

About