Data-driven robust solution schemes for sequential decision making

Park, Hyuk

Permalink

https://hdl.handle.net/2142/132544 Copy

Description

Title

Data-driven robust solution schemes for sequential decision making

Author(s)

Park, Hyuk

Issue Date

2025-12-02

Doctoral Committee Chair(s)

Hanasusanto, Grani Adiwena

Committee Member(s)

• Dayanıklı, Gökçe
• Etesami, Rasoul
• Zhang, Shixuan

Department of Study

Industrial&Enterprise Sys Eng

Discipline

Industrial Engineering

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Stochastic Optimization
• Distributionally Robust Optimization
• Sequential Decision Making
• Multistage Stochastic Programming
• Path Integral Control
• System Identification

Abstract

This dissertation develops robust and data-efficient methodologies for sequential decision making under uncertainty, motivated by challenges arising in operations research, control, and machine learning. Classical approaches such as sample average approximation—also referred to as empirical risk minimization in the machine learning literature—often suffer from poor out-of-sample performance when data is limited. To address this issue, the dissertation proposes a data-efficient alternative to sample average approximation for multistage stochastic programming with Markovian uncertainty and introduces robust and distributionally robust optimization frameworks for two additional problem domains: fairness-aware stochastic optimal control and system identification from a single trajectory. The proposed robust formulations yield tractable optimization problems that can be efficiently solved using off-the-shelf commercial solvers while providing rigorous non-asymptotic performance guarantees. In several chapters, our analysis further uncovers interesting connections between robustification and regularization, the latter being a widely used heuristic in machine learning and control. These contributions advance both the theory and practice of robust learning and control, offering reliable and scalable solutions for data-driven sequential decision making under uncertainty.

Graduation Semester

2025-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/132544

Copyright and License Information

Copyright 2025 Hyuk Park

Owning Collections