C04-1070 In these experiments , we use a one-against-all learning method , which means that we
hide detail