Abstract: Highlights•Proposes a novel SE-shapelets method for semi-supervised time series clustering.•Defines a new SSC method that can avoid un-informative candidate shapelets.•Designs an effective LDS algorithm to discovery representative shapelets.
Loading