Semi-Supervised Learning for Classification

Graph-based and self-training methods for semi-supervised learning

You can use semi-supervised learning techniques when only a small portion of your data is labeled and determining true labels for the rest of the data is expensive. Rather than using a supervised learning method to train a classifier on the labeled data and predict labels for the unlabeled data, you can leverage semi-supervised learning methods to fit labels to the unlabeled data.

If you want to predict labels for new data, you can use the predict object function of the semi-supervised classifier trained on both the labeled and unlabeled data.

Functions

`fitsemigraph`	Label data using semi-supervised graph-based method (Since R2020b)
`fitsemiself`	Label data using semi-supervised self-training method (Since R2020b)
`predict`	Label new data using semi-supervised graph-based classifier (Since R2020b)
`predict`	Label new data using semi-supervised self-trained classifier (Since R2020b)

Objects

`SemiSupervisedGraphModel`	Semi-supervised graph-based model for classification (Since R2020b)
`SemiSupervisedSelfTrainingModel`	Semi-supervised self-trained model for classification (Since R2020b)

Topics

Label Data Using Semi-Supervised Learning Techniques
Compare graph-based and self-training semi-supervised learning techniques.