Hybrid Neural Network/Parametric Models for State and Trait Estimation in Performance Modeling
Agency / Branch:
DOD / ARMY
In sustained military operations involving round-the-clock missions and travel around the globe, biomathematical models of human performance have potential as tools for predicting the fatigue and performance of sleep-deprived soldiers. This project investigates hybrid neural network/parametric models to address specific gaps in current performance models, identified in the 2002 Fatigue and Performance Modeling Workshop and outlined in the Program Solicitation. This project will follow a conventional life-cycle development protocol where models are systematically developed using novel hybrid techniques, validated, and refined to produce an operationally relevant model. The project's main deliverable will be a proposed algorithm for hybrid neural network/parametric performance prediction in the field given unknown individual trait characteristics and uncertain current state, and a comparative analysis to parametric algorithms.
Small Business Information at Submission:
Research Institution Information:
Pulsar Informatics, Inc.
1 Old Dominion University Norfolk, VA 23529
Number of Employees:
1 Old Dominion University
Norfolk, VA 23529
Domestic nonprofit research organization