Predicting Performance from Network Data

Abstract

This project focuses on developing new machine learning and artificial intelligence methods for obtaining valuable inferences from network-structured data. In particular, a focus is on studying team dynamics and how it relates to performance of a team, both in isolation and in adversarial contexts in competing against rival teams or enemies. Most methods for evaluating an individualÕs traits and abilities focus on performance assessments that are conducted on that individual in isolation; for example, having them take an IQ or other assessment test. However, the ability to thrive in a team context and contribute in a positive manner to team performance is not necessarily accurately predicted by isolated assessment tests and individual-specific performance statistics. With this motivation, we focus on obtaining more accurate methods for measuring how each individual contributes to the team based on dynamically collected data on interactions among the team members and how the team performs. The multi-faceted goal is to obtain more nuanced assessments of individual performance based on their contribution to the team, while also developing better methods for predicting the teamÕs performance in different contexts, and for formulating teams that should have better performance than would be predicted from individual assessments. In order to accomplish this goal, we focus on developing significantly innovative models that generalize existing approaches for characterizing social network data to allow the individualÕs in the network to be organized into multiple teams that compete against each other in a spatial domain over time. There is a vast literature on social network data analysis, but the vast majority of the literature is on applying overly-simple models to a single network of individuals, not allowing the dynamic and spatial and interacting nature of realistic team settings. This project will develop significantly new and useful models and corresponding machine learning and AI tools.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 14, 2019
Source ID
W911NF1610544

Entities

People

  • David B. Dunson

Organizations

  • Army Contracting Command
  • Duke University
  • United States Army

Tags

Fields of Study

  • Computer science

Readers

  • Computational Modeling and Simulation
  • Systems Analysis and Design
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks