Improve Model Performance Using Semi-Supervised Wrapper Methods

Hands-on guide to implement and validate self-training using Python

Semi-Supervised Learning is an actively researched field in the machine learning community. It is typically used in improving the generalizability of a supervised learning problem (i.e. training a model based on provided input and ground-truth or actual output value per observation) by leveraging high volumes of unlabeled data (i.e. observations for which inputs or features are available but a ground truth or actual output value …

multilayer-perceptron nlp performance self-training semi-supervised semi-supervised learning unlabeled-data wrapper