Abstract: This paper presents a probabilistic framework for discrepancy prediction in dynamical system models under untested input time histories, based on information gained from validation experiments. Two surrogate modeling-based ...