Semi-supervised learning is a class of machine learning tasks and techniques that make use of a small amount of labeled data together with a large amount of unlabeled data for training. In this report, we focus on one of the techniques, namely, Manifold regularization, which leverages the shape of a dataset to constrain the functions that should be learned on that dataset. We mainly discuss two learning algorithms using manifold regularization, LapRLS and LapSVM. By evaluating on our dataset and comparing with the results of K-means and SVM, we present experimental evidence suggesting that these semi-supervised learning algorithms using manifold regularization can take advantage of both labeled and unlabeled data effectively and perform better than supervised or unsupervised models.
contact hengjia@umich.edu if you have any question