Figure 1From: Protein Ranking by Semi-Supervised Network PropagationAn example of the use of semi-supervised classification. (A) If only one example of each class is given (the large cross and circle), then one may choose an incorrect decision boundary between the two classes. (B) Given extra unlabeled data (small crosses), the decision boundary can be placed in a sparse region of the space, which can result in a more accurate classifier.Back to article page