Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures

Yiliang Guo; Osama Waqar Bhatti; Madhavan Swaminathan

doi:10.1109/EDAPS56906.2022.9994897

Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures

Yiliang Guo, Osama Waqar Bhatti, Madhavan Swaminathan

Electrical Engineering

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

Abstract

Neural Networks surrogate modeling for EM simulations saves computational and design time. Introducing uncertainty estimates into deterministic prediction models provides insight into the reliability and confidence of the model. However, gathering training data to train models is a very time-consuming and resource-consuming task. In this paper, we introduce a method to harness useful insights from confidence bounds to reduce the training set size required to train a model with reasonable accuracy and latency. Using a high-speed differential via structure, we show that the training samples required are 35% less with a slight trade-off in accuracy using the proposed method.

Original language	English (US)
Title of host publication	2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781665491945
DOIs	https://doi.org/10.1109/EDAPS56906.2022.9994897
State	Published - 2022
Event	2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022 - Urbana, United States Duration: Dec 12 2022 → Dec 14 2022

Publication series

Name	IEEE Electrical Design of Advanced Packaging and Systems Symposium
Volume	2022-December
ISSN (Print)	2151-1225
ISSN (Electronic)	2151-1233

Conference

Conference	2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022
Country/Territory	United States
City	Urbana
Period	12/12/22 → 12/14/22

All Science Journal Classification (ASJC) codes

Control and Systems Engineering
Electrical and Electronic Engineering
Automotive Engineering
Computer Science(all)

Access to Document

10.1109/EDAPS56906.2022.9994897

Cite this

Guo, Y., Bhatti, O. W., & Swaminathan, M. (2022). Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures. In 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022 (IEEE Electrical Design of Advanced Packaging and Systems Symposium; Vol. 2022-December). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/EDAPS56906.2022.9994897

Guo, Yiliang ; Bhatti, Osama Waqar ; Swaminathan, Madhavan. / Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures. 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022. Institute of Electrical and Electronics Engineers Inc., 2022. (IEEE Electrical Design of Advanced Packaging and Systems Symposium).

@inproceedings{a69ee7778d9546018d4dd3eef19dbbe1,

title = "Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures",

abstract = "Neural Networks surrogate modeling for EM simulations saves computational and design time. Introducing uncertainty estimates into deterministic prediction models provides insight into the reliability and confidence of the model. However, gathering training data to train models is a very time-consuming and resource-consuming task. In this paper, we introduce a method to harness useful insights from confidence bounds to reduce the training set size required to train a model with reasonable accuracy and latency. Using a high-speed differential via structure, we show that the training samples required are 35% less with a slight trade-off in accuracy using the proposed method.",

author = "Yiliang Guo and Bhatti, {Osama Waqar} and Madhavan Swaminathan",

note = "Funding Information: This work was supported by DARPA under the Warden program (Project Number GR00013386). Publisher Copyright: {\textcopyright} 2022 IEEE.; 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022 ; Conference date: 12-12-2022 Through 14-12-2022",

year = "2022",

doi = "10.1109/EDAPS56906.2022.9994897",

language = "English (US)",

series = "IEEE Electrical Design of Advanced Packaging and Systems Symposium",

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

booktitle = "2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022",

address = "United States",

}

Guo, Y, Bhatti, OW & Swaminathan, M 2022, Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures. in 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022. IEEE Electrical Design of Advanced Packaging and Systems Symposium, vol. 2022-December, Institute of Electrical and Electronics Engineers Inc., 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022, Urbana, United States, 12/12/22. https://doi.org/10.1109/EDAPS56906.2022.9994897

Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures. / Guo, Yiliang; Bhatti, Osama Waqar; Swaminathan, Madhavan.
2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022. Institute of Electrical and Electronics Engineers Inc., 2022. (IEEE Electrical Design of Advanced Packaging and Systems Symposium; Vol. 2022-December).

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

TY - GEN

T1 - Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures

AU - Guo, Yiliang

AU - Bhatti, Osama Waqar

AU - Swaminathan, Madhavan

PY - 2022

Y1 - 2022

N2 - Neural Networks surrogate modeling for EM simulations saves computational and design time. Introducing uncertainty estimates into deterministic prediction models provides insight into the reliability and confidence of the model. However, gathering training data to train models is a very time-consuming and resource-consuming task. In this paper, we introduce a method to harness useful insights from confidence bounds to reduce the training set size required to train a model with reasonable accuracy and latency. Using a high-speed differential via structure, we show that the training samples required are 35% less with a slight trade-off in accuracy using the proposed method.

AB - Neural Networks surrogate modeling for EM simulations saves computational and design time. Introducing uncertainty estimates into deterministic prediction models provides insight into the reliability and confidence of the model. However, gathering training data to train models is a very time-consuming and resource-consuming task. In this paper, we introduce a method to harness useful insights from confidence bounds to reduce the training set size required to train a model with reasonable accuracy and latency. Using a high-speed differential via structure, we show that the training samples required are 35% less with a slight trade-off in accuracy using the proposed method.

UR - http://www.scopus.com/inward/record.url?scp=85146146228&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85146146228&partnerID=8YFLogxK

U2 - 10.1109/EDAPS56906.2022.9994897

DO - 10.1109/EDAPS56906.2022.9994897

M3 - Conference contribution

AN - SCOPUS:85146146228

T3 - IEEE Electrical Design of Advanced Packaging and Systems Symposium

BT - 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022

Y2 - 12 December 2022 through 14 December 2022

ER -

Guo Y, Bhatti OW, Swaminathan M. Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures. In 2022 IEEE Electrical Design of Advanced Packaging and Systems, EDAPS 2022. Institute of Electrical and Electronics Engineers Inc. 2022. (IEEE Electrical Design of Advanced Packaging and Systems Symposium). doi: 10.1109/EDAPS56906.2022.9994897

Training Set Optimization with Uncertainty Quantification for Machine Learning Models of Electromagnetic Structures

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this