We view teams as semantically-labeled, multilevel networks with attributes on nodes, edges, and subgroups. Through experiments for validation, we develop models of team performance: both statistical models using networks and fine-grain models of information flows. We then examine the relationship between influence networks and collective intelligence on task performance, as we consider the robustness and optimization of teams.

We  examine the performance of teams on a sequence of tasks. The first part of this research examines the changes in social influence networks resulting from repeated tasks. The second part explores team adaptivity using methods from computational learning and data-driven optimization.

We examine scalability of teams and tasks by considering complex tasks that require a team-of-teams approach. Given a complex task, this thrust considers how to decompose it into subtasks, the effect of inter-task dependencies, and how to assign teams to such tasks. We consider the learning of optimal decomposition and coordination structures using methods from reinforcement learning, and we examine adaptivity over nested tasks and learning of optimal coordination structures.