Amir MOHAMMADI
--- a/bob/fusion/base/script/bob_fuse.py
+++ b/bob/fusion/base/script/bob_fuse.py
    if args.eval_files:
      check_consistency(gen_le, zei_le, atk_le)

-  scores_train = get_scores(score_lines_list_train)
+  scores_train = get_scores(gen_lt, zei_lt, atk_lt)
+  # scores_train = get_scores(score_lines_list_train)
  train_neg = get_scores(zei_lt, atk_lt)
--- a/bob/fusion/base/script/bob_fuse.py
+++ b/bob/fusion/base/script/bob_fuse.py
      score_lines['score'] = fused_scores_eval
      create_directories_safe(os.path.dirname(args.fused_eval_file))
      dump_score(args.fused_eval_file, score_lines)
+      # to separate scores for licit and spoof scenarios
+      start_zei = len(gen_le[0])
+      start_atk = start_zei + len(zei_le[0])
+      dump_score(args.fused_eval_file + '-licit', score_lines[0:start_atk])
+      import numpy
+      dump_score(args.fused_eval_file + '-spoof', numpy.append(score_lines[0:start_zei], score_lines[start_atk:-1]))
--- a/bob/fusion/base/script/plot_fusion_decision_boundary.py
+++ b/bob/fusion/base/script/plot_fusion_decision_boundary.py
+    bob.core.log.add_command_line_option(parser)
+
+    # parse command line options
+    args = parser.parse_args(command_line_parameters)
+    bob.core.log.set_verbosity_level(logger, args.verbose)
+
+    # load the algorithm
+    algorithm = bob.fusion.base.algorithm.Algorithm()
+    algorithm = algorithm.load(args.model_file)
+
+    hdf5_path = os.path.splitext(args.model_file)[0] + '.hdf5'
+    machine = None
+    # if os.path.isfile(hdf5_path):
+    #     hdf5file = bob.io.base.HDF5File(hdf5_path)
+    #     machine = bob.learn.linear.Machine(hdf5file)
+
--- a/bob/fusion/base/script/plot_fusion_decision_boundary.py
+++ b/bob/fusion/base/script/plot_fusion_decision_boundary.py
+    # plot the decision boundary
+    do_grouping = True
+    if args.group < 1:
+        do_grouping = False
+
+    plot_boundary_decision(
+        algorithm, scores, score_labels, args.threshold,
+        do_grouping=do_grouping,
+        npoints=args.group,
+        seed=0,
+        gformat=args.grouping
+    )
+    plt.savefig(args.output)
+    plt.close()
+
+    if machine is not None:
--- a/bob/fusion/base/algorithm/CascadeFuse.py 0 → 100644
+++ b/bob/fusion/base/algorithm/CascadeFuse.py 0 → 100644
+
+class CascadeFuse(Algorithm):
+  """weighted sum (default: mean)"""
+
+  def __init__(self, *args, **kwargs):
+    super(CascadeFuse, self).__init__(
+      classifier=self,
+      *args, **kwargs)
+
+  def fit(self, X, y):
+    # no training for this approach
+    pass
+
+  def decision_function(self, scores):
+    # we assume that scores in 1st column are used to decide outcome
+    return [score_set[0] if score_set[0] < 0.2 else score_set[1] for score_set in scores]