Physics Colloquium: Hybrid Forecasting of Complex Systems: Combining Machine Learning with Knowledge-based Models
Prof. Michelle Girvan, University of Maryland
In recent years, machine learning methods such as "deep learning" have proven enormously successful for tasks such as image classification, voice recognition, and more. Despite their effectiveness for big-data classification problems, these methods have had limited success for time series prediction, especially for complex systems like those we see in weather, solar activity, and brain dynamics. In this talk, I will discuss how a Reservoir Computer (RC) - a special kind of machine learning system that offers a "universal" dynamical system - can draw on its own internal complex dynamics in order to forecast systems like the weather, beyond the time horizon of other methods. The RC provides a knowledge-free approach because it builds forecasts purely from past measurements without any specific knowledge of the system dynamics. By building a new hybrid approach that judiciously combines the knowledge-free prediction of the RC with a knowledge-based, mechanistic model, we demonstrate a further, dramatic, improvement in forecasting complex systems. This hybrid approach can given us new insights into the weaknesses of our knowledge-based models and also reveal limitations in our machine learning system, guiding improvements in both knowledge-free and knowledge-based prediction techniques.
Tuesday, April 2, 2019 at 3:15pm
Regents Hall, 109
3700 O St. NW