Title: Navigating Complex Simulations: The Power of Visual Exploration
Presentor: Dr. Kresimir Matkovic (Senior Researcher, VRVis GmbH, Vienna Research Center for Visual Computing, Austria)
Abstract.
High-dimensional parameter spaces in modern simulations pose significant challenges for analysis and interpretation. While interactive visual analysis excels at revealing input-output relations, its effectiveness can diminish with increasing model complexity or atypical output characteristics. Integrating automated techniques, such as deep learning, with interactive exploration offers a robust solution in many cases. In this talk, we present two use cases highlighting different approaches: (1) a cooling system simulation, where a deep-learning-based inverse model is combined with expert-driven interaction design to support engineering workflows; and (2) parameter calibration in complex social simulations, specifically epidemiological models, where advanced interactive visual techniques enable users to effectively explore and interpret emergent behaviors.
