Python codes for weakly-supervised learning

Simple sklearn based python implementation of Positive-Unlabeled (PU) classification using bagging based ensembles

Positive and Unlabeled Materials Machine Learning (pumml) is a code that uses semi-supervised machine learning to classify materials from only positive and unlabeled examples.

A collection of notebooks that implement algorithms introduced in "Learning from positive and unlabeled data: a survey"

A curated list of resources dedicated to Positive Unlabeled(PU) learning ML methods.

Python framework for interpretable protein prediction

An example repo for how PU Bagging and TSA works.

🍊 PAUSE (Positive and Annealed Unlabeled Sentence Embedding), accepted by EMNLP'2021 🌴

uPU, nnPU and PN learning with Extra Trees classifier.

NeurIPS'20 Paper: "Learning from Positive and Unlabeled Data with Arbitrary Positive Shift"

Predicting protein functions using positive-unlabeled ranking with ontology-based priors

A template for a PU Bagging approach. PU bagging is effective when reliable negatives can't be identified in unlabeled data. Bootstrapping creates resampled subsets, helping the model distinguish true positives from true negatives. This process infers the negative class distribution, improving classification and model robustness.

PyTorch Implementation of Asymetric Cross Entropy Loss (Loss Correction for PU Learning)

Domain Adaptation with Dynamic Open-Set Targets