Machine learning surrogate modeling methods in inverse high-speed link design

This thesis implements and compares the performance of several Machine Learning surrogate modeling methods for the inverse high-speed channel design problem. A Tandem Neural Network structure and a User-Choice Inverse Neural Network structure are purposed and thoroughly described for the inverse optimization of high-speed link problems. Comparisons are made between the newly purposed methods and the traditional Machine Learning methods. There are discussions on inverse optimization with mixed continuous-discrete variables. Three high-speed link designs are constructed as examples to evaluate the performance of the Machine Learning methods. Different types of error calculation are displayed to better compare the prediction results of different Machine Learning methods.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical & Computer Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2022

Author dc:creator

• Wang, Yuechen

Contributors dc:contributor

• Chen, Xu

• Signal Integrity
• Inverse Problem
• Machine Learning
• Neural Network
• High-speed Link Design

Statement dc:rights

• Copyright 2022 Yuechen Wang

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/117692

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F117692