Simulation is a useful and effective way to analyze and study complex, real-world systems. It allows researchers, practitioners, and decision makers to make sense of the inner working of a system that involves many factors often resulting in some sort of emerging behavior. Scenarios such as the spread of a pandemic, the operations of an autonomous vehicle on busy streets, or the flow of patients in an emergency room can be studied with simulation models. Agent based modeling or ABM is a common modeling technique used in simulating and studying such complex systems. In these models, agents are individual autonomous entities that make decisions about their actions and interactions within the environment. The factors that influence the agent's decision making process and thus drive the simulation outcome are commonly known as parameters. A typical agent-based simulation model will include many parameters, each with a potentially large set of values. The number of scenarios with different parameter value combinations grows exponentially and quickly becomes infeasible to test them all or even to explore a suitable subset of them. How does one then efficiently identify the parameter value combinations that matter for a particular simulation study? In addition, is it possible to train a machine learning model to predict the outcome of an agent-based model without running the agent-based model for all parameter value combinations? In this presentation, we discuss the problem of handling large parameter spaces, the use of covering arrays to decrease the space for this type of problem, and the application of machine learning to predict the result of ABMs using the covering array data to choose a representative part of the parameter value space.
Simulation is a useful and effective way to analyze and study complex, real-world systems. It allows researchers, practitioners, and decision makers to make sense of the inner working of a system that involves many factors often resulting in some sort of emerging behavior. Scenarios such as the...
See full abstract
Simulation is a useful and effective way to analyze and study complex, real-world systems. It allows researchers, practitioners, and decision makers to make sense of the inner working of a system that involves many factors often resulting in some sort of emerging behavior. Scenarios such as the spread of a pandemic, the operations of an autonomous vehicle on busy streets, or the flow of patients in an emergency room can be studied with simulation models. Agent based modeling or ABM is a common modeling technique used in simulating and studying such complex systems. In these models, agents are individual autonomous entities that make decisions about their actions and interactions within the environment. The factors that influence the agent's decision making process and thus drive the simulation outcome are commonly known as parameters. A typical agent-based simulation model will include many parameters, each with a potentially large set of values. The number of scenarios with different parameter value combinations grows exponentially and quickly becomes infeasible to test them all or even to explore a suitable subset of them. How does one then efficiently identify the parameter value combinations that matter for a particular simulation study? In addition, is it possible to train a machine learning model to predict the outcome of an agent-based model without running the agent-based model for all parameter value combinations? In this presentation, we discuss the problem of handling large parameter spaces, the use of covering arrays to decrease the space for this type of problem, and the application of machine learning to predict the result of ABMs using the covering array data to choose a representative part of the parameter value space.
Hide full abstract
