Datasets and Methods

The labeled and unlabeled dataset used to test the HEAL annotation server is available below.

The data consist of 9 features with numeric data type.

HEAL provides a reliable and costless alternative to existing Active learning that requires an human-in-the-loop to reinforce the accuracy of base model.