Interpretable machine learning models: a physics-based view

Ion Matei, Johan de Kleer, Christoforos Somarakis +2

To understand changes in physical systems and facilitate decisions, explaining how model predictions are made is crucial. We use model-based interpretability, where models of physi…