Title: Data-Driven Decision Analytics in Complex Systems: Problem Decomposition and Algorithmic Design

Date: 3/16/2026

Time: 11:30AM - 12:30AM

Location: Groseclose 303 Conference Room, 765 Ferst Dr NW, Atlanta, GA 30332

Zoom Link: https://gatech.zoom.us/j/97669507246

Yujia Xie

Machine Learning PhD Student

H. Milton Stewart School of Industrial and Systems Engineering

Georgia Institute of Technology

Thesis Committee

Dr. Nicoleta Serban (advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Gian-Gabriel Garcia, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Shihao Yang, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Mahdi Noorizadegan, Newcastle Business School, Northumbria University

Dr. Jovan Julien, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Eunhye Song, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Abstract

Effective decision-making in complex systems, such as healthcare and logistics, requires balancing high-fidelity modeling with the computational challenges of large-scale, high-dimensional environments. This thesis develops comprehensive computational frameworks for solving these problems by integrating optimization, simulation, and machine learning methodologies.

The research first applies these integrated techniques to a critical real-world case: improving psychosocial healthcare access for Medicaid-insured children through network-based decision insights. To address scalability, the thesis leverages structural clustering and decomposition methods to facilitate load-balanced parallel computation. For example, Lagrangian and Dantzig-Wolfe in optimization, as well as spatial, agent-based, and functional decompositions in simulation. Furthermore, the integration of machine learning surrogate models replaces computationally expensive subroutines, providing algorithmic acceleration.