WIZER: An Automated Intelligent Tool For Model Improvement Of Simulations

Alex Yahja, Computer Science, Carnegie Mellon University

As simulations increase in fidelity and become a more useful tool for science, simulation models grow in size and complexity which requires significant time and effort to validate and improve the models. I have developed a knowledge-based simulation-aided and simulation-control approach for automating model-validation and model-improvement of simulation. The knowledge and assumptions of the simulation are examined through inference and simulation. Hypotheses are constructed and tested with the help of artificial intelligence techniques. A tool implementing the above approach (WIZER) has been constructed and tested on agent-based simulation models. I present some results of WIZER.