Date of Award

Spring 2026

Document Type

Thesis (Ph.D.)

Department or Program

Engineering Sciences

First Advisor

Eugene Santos Jr

Abstract

Across human domains ranging from sports to business and organizational settings, complex tasks are often solved by teams rather than individuals, leveraging benefits such as interaction, mutual support, complementary skills, cohesion, and task allocation. Evaluating team effectiveness, however, is inherently challenging due to the subjectivity of many existing techniques and the limitations of outcome-driven metrics that primarily focus on performance scores while overlooking the team processes that generated the scores. To address these challenges, this dissertation proposes a behavioral-centric, end-to-end framework for team evaluation grounded in reward functions that model sequential team behavior. Reward functions offer compact and interpretable representations of team preferences and objectives and can be inferred via Inverse Reinforcement Learning (IRL) from observed behavior. Unlike manually hand-crafted rewards, IRL-derived rewards capture nuanced and implicit team goals embedded within historical behaviors. However, these learned reward functions may suffer from misspecification arising from data sparsity and training instabilities. To enhance robustness and interpretability, this dissertation introduces the concept of reward consistency as a formal measure of the stability and reliability of learned reward functions under similar preferential conditions. In addition, it explores canonicalization techniques as preprocessing methods to standardize learned reward functions, thereby mitigating misspecification to improve the efficiency of downstream tasks such as behavioral classification and Reinforcement Learning (RL). The refined reward representations enable richer forms of team evaluation, which we segment into three key tasks: behavioral attribute extraction, team-impact rankings, and performance prediction.

Download

Share

COinS