TABLE I.
Method | Multiple classifiers | Identify easy/hard | Unlabeled data | Distribution mismatch | Smoothness assumption | Non-parallel hyperplanes | Progressive labeling |
---|---|---|---|---|---|---|---|
Boosting [27] | ✓ | × | × | × | × | × | × |
TW-SVM [37] | ✓ | × | × | × | × | ✓ | × |
Self-paced learning [39] | × | ✓ | × | × | × | × | × |
RO-SVM [4], [33] | × | ✓ | × | × | × | × | × |
Self-training [46] | × | × | ✓ | × | × | × | ✓ |
Co-training [6] | ✓ | × | ✓ | × | × | × | ✓ |
Lap-SVM [43] | × | × | ✓ | × | ✓ | × | × |
DAM [21] | ✓ | × | ✓ | ✓ | ✓ | × | ✓ |
CPM (proposed) | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |