I am working on a new idea to improve the classification accuracy in semi-supervised learning. I want to use the same text classification dataset and split this dataset into labeled set and unlabeled set, How I can do that in Java?
Can anyone help me?
It is not going to give you an improved accuracy when using less labels. If you split you data in order to delete the labels from one group and use that with a semi-supervised learning, it will just reduce your accuracy. The purpose of semi-supervised is that the process of labeling massive amounts of data for supervised learning is extremely time-consuming and expensive, so if you need more data (that you already have), then you can use techniques to use unlabeled data. Before even thinking about the coding in Java, can you develop a bit more your idea of why you were thinking of this?