Modelling Unsupervised Event Segmentation: Learning Event Boundaries from Prediction Errors | AMiner