PhD Dissertation Proposal: Shuo Liu
Title: “Redundancy in Resilient Distributed Optimization and Related Problems”
In this dissertation proposal, we study various resilient distributed optimization and related problems, focusing on the importance of redundancy in the inputs in solving the problems.
We begin with the existing work on Byzantine aggregate optimization under a server-based architecture that can be solved exactly, including the Byzantine set intersection problem in various settings and the Byzantine aggregate optimization problem under decentralized architecture. We then proposed and discussed the approximate Byzantine aggregate optimization problem. We further extend this problem to the approximate resilient aggregate optimization problem against both Byzantine agents and stragglers. We also propose and examine the Byzantine min-max optimization problem.
For the problems discussed, we utilize redundancy in both showing solvabilities and analyzing practical algorithms.
Committee members:
Nitin Vaidya (adviser)
Calvin Newport
Lisa Singh
Thinh Doan (Virginia Tech)