Behavior
Modeling and simulation for Behavior include both descriptive and prescriptive models. Descriptive models attempt to characterize behavioral processes—such as decision making—in ways that use a systematic description of behaviors and their influences and typically provide general predictions. Prescriptive models use computational, numerical, or simulation-based methods to represent behavioral processes more formally and define a baseline for measuring human behavior.
Models developed at the Pacific Northwest National Laboratory apply a wide range of science and methods to predict various aspects of individual actions and social interactions.
- Social, behavioral, and economic science can be used to model scenarios of interest, to develop relevant collections of current and historical data as sources of evidence, and to engage Bayesian Networks and Dempster-Schaefer algorithms to estimate the likelihood of plausible outcomes in these scenarios.
- Probabilistic modeling, declarative ontological modeling, and object-oriented simulation can be used to express human, societal, and hardware behavior and assess their impacts on systems.
- Ontological models may be based on careful examination of a dataset containing the behavioral patterns thought to be important to the model or based on Subject Matter Expert (SME) knowledge of the behaviors to be modeled.
- System Dynamics Models are used to help characterize and understand relationships among variables or constructs.
- Artificial Neural Networks and other machine learning methods are used to model behavior of complex adaptive systems in diverse applications such as pattern recognition, health monitoring, and learning systems.
- Business process re-engineering can be evaluated through modeling. Developments include computational representations of social sciences influencing decision making, modeling uncertainties of clustering memberships for estimating uncertainties in fused clustering solutions, modeling semantic relationships between ontological concepts, and modeling fusion strategies to enable their formal comparison.