We address the problem of feature weight learning for image clustering. In practice, before clustering data, we generally normalize all data features between 0 and 1, because we cannot determine which ...