Confidence Estimation for Graph-based Semi-supervised Learning

To select unlabeled example effectively and reduce classification error, confidence estimation for graphbased semi-supervised learning (CEGSL) is proposed. This algorithm combines graph-based semi-supervised learning with collaboration-training. It makes use of structure information of sample to calculate the classification probability of unlabeled example explicitly. With multi-classifiers, the algorithm computes the confidence of unlabeled example implicitly. With dualconfidence estimation, the unlabeled example is selected to update classifiers. The comparative experiments on UCI datasets indicate that CEGSL can effectively exploit unlabeled data to enhance the learning performance.