Channel Inverse Design Using Tandem Neural Network

Hanzhi Ma; Er Ping Li; Yuechen Wang; Bobi Shi; Jose Schutt-Aine; Andreas Cangellaris; Xu Chen

doi:10.1109/SPI54345.2022.9874935

Channel Inverse Design Using Tandem Neural Network

Hanzhi Ma, Er Ping Li, Yuechen Wang, Bobi Shi, Jose Schutt-Aine, Andreas Cangellaris, Xu Chen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

A tandem neural network (NN) with R2 score-based loss function is proposed in this paper for channel inverse design. Tandem NN consists of an inverse neural network from target performance to design parameters and a pre-trained forward neural network from design parameters to design targets. The training of the actual INN uses the fixed pre-trained forward model to evaluate the inverse design output. A channel inverse design example for target impedance and attenuation at multiple frequency points is applied in this paper to evaluate the performance of tandem NN. Numerical results show that tandem NN achieves a good design result compared with target performance and regular NN.

Original language	English (US)
Title of host publication	26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781665486255
DOIs	https://doi.org/10.1109/SPI54345.2022.9874935
State	Published - 2022
Event	26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Siegen, Germany Duration: May 22 2022 → May 25 2022

Publication series

Name	26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings

Conference

Conference	26th IEEE Workshop on Signal and Power Integrity, SPI 2022
Country/Territory	Germany
City	Siegen
Period	5/22/22 → 5/25/22

Keywords

high-speed link
impedance and attenuation
inverse design
neural network

ASJC Scopus subject areas

Energy Engineering and Power Technology
Nuclear Energy and Engineering
Safety, Risk, Reliability and Quality
Control and Optimization
Computer Networks and Communications
Signal Processing

Online availability

10.1109/SPI54345.2022.9874935

Library availability

Discover UIUC Full Text

Cite this

Ma, H., Li, E. P., Wang, Y., Shi, B., Schutt-Aine, J., Cangellaris, A., & Chen, X. (2022). Channel Inverse Design Using Tandem Neural Network. In 26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings (26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/SPI54345.2022.9874935

Channel Inverse Design Using Tandem Neural Network. / Ma, Hanzhi; Li, Er Ping; Wang, Yuechen et al.
26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings. Institute of Electrical and Electronics Engineers Inc., 2022. (26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Ma, H, Li, EP, Wang, Y, Shi, B, Schutt-Aine, J , Cangellaris, A & Chen, X 2022, Channel Inverse Design Using Tandem Neural Network. in 26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings. 26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings, Institute of Electrical and Electronics Engineers Inc., 26th IEEE Workshop on Signal and Power Integrity, SPI 2022, Siegen, Germany, 5/22/22. https://doi.org/10.1109/SPI54345.2022.9874935

Ma H, Li EP, Wang Y, Shi B, Schutt-Aine J , Cangellaris A et al. Channel Inverse Design Using Tandem Neural Network. In 26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings. Institute of Electrical and Electronics Engineers Inc. 2022. (26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings). doi: 10.1109/SPI54345.2022.9874935

@inproceedings{ba223c2e18d949d6a82e2a709d77a258,

title = "Channel Inverse Design Using Tandem Neural Network",

abstract = "A tandem neural network (NN) with R2 score-based loss function is proposed in this paper for channel inverse design. Tandem NN consists of an inverse neural network from target performance to design parameters and a pre-trained forward neural network from design parameters to design targets. The training of the actual INN uses the fixed pre-trained forward model to evaluate the inverse design output. A channel inverse design example for target impedance and attenuation at multiple frequency points is applied in this paper to evaluate the performance of tandem NN. Numerical results show that tandem NN achieves a good design result compared with target performance and regular NN.",

keywords = "high-speed link, impedance and attenuation, inverse design, neural network",

author = "Hanzhi Ma and Li, {Er Ping} and Yuechen Wang and Bobi Shi and Jose Schutt-Aine and Andreas Cangellaris and Xu Chen",

note = "ACKNOWLEDGMENT This work of Hanzhi Ma and Er-Ping Li were sponsored by the National Natural Science Foundation of China under Grant No. 62071424 and 62027805, Zhejiang Laboratory Foundation of China under Grant No. 2020KCDAB01, and Zhejiang Provincial Natural Science Foundation of China under Grant No. LD21F010002. Hanzhi Ma was supported by Zhejiang University Academic Award for Outstanding Doctoral Candidates. Yuechen Wang and Xu Chen were supported by the National Science Foundation under Grant No. CNS 16-24811 - Center for Advanced Electronics through Machine Learning (CAEML).; 26th IEEE Workshop on Signal and Power Integrity, SPI 2022 ; Conference date: 22-05-2022 Through 25-05-2022",

year = "2022",

doi = "10.1109/SPI54345.2022.9874935",

language = "English (US)",

series = "26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings",

address = "United States",

}

TY - GEN

T1 - Channel Inverse Design Using Tandem Neural Network

AU - Ma, Hanzhi

AU - Li, Er Ping

AU - Wang, Yuechen

AU - Shi, Bobi

AU - Schutt-Aine, Jose

AU - Cangellaris, Andreas

AU - Chen, Xu

N1 - ACKNOWLEDGMENT This work of Hanzhi Ma and Er-Ping Li were sponsored by the National Natural Science Foundation of China under Grant No. 62071424 and 62027805, Zhejiang Laboratory Foundation of China under Grant No. 2020KCDAB01, and Zhejiang Provincial Natural Science Foundation of China under Grant No. LD21F010002. Hanzhi Ma was supported by Zhejiang University Academic Award for Outstanding Doctoral Candidates. Yuechen Wang and Xu Chen were supported by the National Science Foundation under Grant No. CNS 16-24811 - Center for Advanced Electronics through Machine Learning (CAEML).

PY - 2022

Y1 - 2022

N2 - A tandem neural network (NN) with R2 score-based loss function is proposed in this paper for channel inverse design. Tandem NN consists of an inverse neural network from target performance to design parameters and a pre-trained forward neural network from design parameters to design targets. The training of the actual INN uses the fixed pre-trained forward model to evaluate the inverse design output. A channel inverse design example for target impedance and attenuation at multiple frequency points is applied in this paper to evaluate the performance of tandem NN. Numerical results show that tandem NN achieves a good design result compared with target performance and regular NN.

AB - A tandem neural network (NN) with R2 score-based loss function is proposed in this paper for channel inverse design. Tandem NN consists of an inverse neural network from target performance to design parameters and a pre-trained forward neural network from design parameters to design targets. The training of the actual INN uses the fixed pre-trained forward model to evaluate the inverse design output. A channel inverse design example for target impedance and attenuation at multiple frequency points is applied in this paper to evaluate the performance of tandem NN. Numerical results show that tandem NN achieves a good design result compared with target performance and regular NN.

KW - high-speed link

KW - impedance and attenuation

KW - inverse design

KW - neural network

UR - http://www.scopus.com/inward/record.url?scp=85138493714&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85138493714&partnerID=8YFLogxK

U2 - 10.1109/SPI54345.2022.9874935

DO - 10.1109/SPI54345.2022.9874935

M3 - Conference contribution

AN - SCOPUS:85138493714

T3 - 26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings

BT - 26th IEEE Workshop on Signal and Power Integrity, SPI 2022 - Conference Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 26th IEEE Workshop on Signal and Power Integrity, SPI 2022

Y2 - 22 May 2022 through 25 May 2022

ER -

Channel Inverse Design Using Tandem Neural Network

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Online availability

Library availability

Related links

Fingerprint

Cite this