Double data piling leads to perfect classification