TM (Two Models) method for biased positive-unlabeled (PU) learning

This repository contains implementation of TM (Two Models) method for positive-unlabeled learning

Repository structure:

• requirements.txt contains dependencies necessary to run the code (install using pip install -r requirements.txt),
• experiement.py contains examples of usage for benchmark dataset,
• data/ directory contains benchmark dataset,
• putm.py contains code of the proposed method as well as the code of the naive method,
• labelling.py contains a function that allows to generate PU labels based on real data,
• utils.py contains auxilliary functions.