In this work, a perturbation-based model-agnostic explanation method for time series classification is presented. One of the main novelties of the proposed method is that the considered perturbations are interpretable and ...