Monday, September 21, 20153:00 PM - 4:00 PMCNLS Conference Room (TA-3, Bldg 1690)|
Thinking out of the Box: All models are wrong,.. can we make them more useful?
Mathematical models are employed ubiquitously for description, prediction and decision making. In addressing end-goal objectives, great care needs to be devoted to attainment of appropriate balance of inexactness throughout the various stages of the end goal process (e.g. modeling and inversion). Disregard to such considerations, either entails redundant computation or impairment of the overall fidelity of the inversion process. Model reduction is instrumental in trading-off fidelity for computation, yet, in some situations, it is essential to take an opposite perspective, and enhance model fidelity and thereby end-goal output. In this talk, we shall describe the interplay between model reduction and model mis-specification mitigation and provide a generic infrastructure for model re-specification based upon a hybrid first principles and data-driven approach.
Host: Misha Chertkov