Back to search

Georgia Institute of Technology

Rapid Assessment and Uncertainty Quantification for Electrical Responses Using Machine Learning Based Predictions and Extrapolations

Abstract

dc:description.abstract

The increasing complexity of advanced packaging and microwave systems has made accurate and efficient electromagnetic (EM) analysis a critical challenge, as traditional full-wave solvers are computationally expensive and limit rapid design exploration, optimization, and uncertainty quantification. This dissertation develops machine learning–based surrogate modeling and extrapolation frameworks for fast and reliable prediction of EM responses, with a focus on mapping low-dimensional design parameters to high-dimensional frequency-domain behaviors such as S-parameters. The core methodology is built upon spectral transposed convolutional neural networks (S-TCNNs), which are enhanced through several key innovations: an extrapolation framework with range determination to enable controlled predictions beyond the training domain, an adaptive deep kernel learning (DKL) approach that integrates Gaussian process modeling for improved generalization and uncertainty quantification, and trainable activation functions that enhance training stability and convergence. The proposed methods are validated on representative advanced packaging structures over wide frequency ranges, demonstrating substantial reductions in computational cost compared to conventional EM solvers while maintaining high predictive accuracy in both interpolation and extrapolation regimes. In addition, the framework provides insight into uncertainty and parameter sensitivity, supporting more robust and informed design decisions. In summary, this work establishes a scalable and interpretable machine learning framework for rapid EM response prediction, contributing to the advancement of next-generation electronic design automation.

Degree

thesis:*
Name thesis:degree_name
Electrical and Computer Engineering, PhD
Department thesis:degree_department
Electrical and Computer Engineering
Grantor
Georgia Institute of Technology
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Guo, Yiliang
Advisor dc:contributor.advisor
  • Swaminathan, Madhavan
Committee members dc:contributor.committeemember
  • Peterson, Andrew F.
  • Mukhopadhyay, Saibal
  • Hao, Callie
  • Durgin, Gregory
  • Liu, Ling

Rights

Language dc:language.iso
English

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/1853/81782

Chain of custody

source
Harvested from
Georgia Tech
Base URL
repository.gatech.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Guo, Yiliang. Rapid Assessment and Uncertainty Quantification for Electrical Responses Using Machine Learning Based Predictions and Extrapolations. Georgia Institute of Technology, 2026. https://hdl.handle.net/1853/81782