Semi-supervised Learning Based on Semiparametric Regularization
Labeled data are often expensive to obtain since they require the efforts of experienced experts. Meanwhile, the unlabeled data are relatively easy to collect. Semiparametric regularization semi-supervised learning attempts to use the unlabeled data to improve the performance. Experimental comparisons demonstrate that our approach outperforms the state-of-the-art methods in the literature on a variety of classification tasks. Therefore, our approach is a promising technology in the machine learning field.
In order to utilize the unlabeled data, the semiparametric regularization semi-supervised learning approach incorporates the marginal distribution of the data into the supervised learning through exploiting the geometric distribution of the data. This approach allows a family of algorithms to be developed based on various choices of the original RKHS and the loss function. Experimental comparisons demonstrate that the proposed approach outperforms the state-of-the-art methods in the literature on a variety of classification tasks.
Binghamton University RB287