Learning from positive and unlabeled data: a survey

Abstract

Learning from positive and unlabeled data or PU learning is the setting where a learner only has access to positive examples and unlabeled data. The assumption is that the unlabeled data can contain both positive and negative examples. This setting has attracted increasing interest within the machine learning literature as this type of data naturally arises in applications such as medical diagnosis and knowledge base completion. This article provides a survey of the current state of the art in PU learning. It proposes seven key research questions that commonly arise in this field and provides a broad overview of how the field has tried to address them.

There was a typo in section 2.4. The fraction of labeled examples in the single-training-set scenario should be αc, and not αe(x) as was written in the previous version

References in corpus (9)

• Classification with Noisy Labels by Importance Reweighting
• One-Class Classification: Taxonomy of Study and Review of Techniques
• Positive-Unlabeled Learning with Non-Negative Risk Estimator
• A Robust Ensemble Approach to Learn From Positive and Unlabeled Data Using SVM Base Models
• Mixture Proportion Estimation via Kernel Embedding of Distributions
• Estimating the class prior and posterior from noisy positives and unlabeled data
• Class-prior Estimation for Learning from Positive and Unlabeled Data
• Learning with Confident Examples: Rank Pruning for Robust Classification with Noisy Labels
• Building Classifiers to Predict the Start of Glucose-Lowering Pharmacotherapy Using Belgian Health Expenditure Data

Cited by in corpus (39)

• Learning Off-Road Terrain Traversability with Self-Supervisions Only
• ScaTE: A Scalable Framework for Self-Supervised Traversability Estimation in Unstructured Environments
• Positive-Unlabeled Learning with Adversarial Data Augmentation for Knowledge Graph Completion
• On the selection and effectiveness of pseudo-absences for species distribution modeling with deep learning
• On Missing Labels, Long-tails and Propensities in Extreme Multi-label Classification
• Ranking the Synthesizability of Hypothetical Zeolites with the Sorting Hat
• NIAPU: network-informed adaptive positive-unlabeled learning for disease gene identification
• From Weakly Supervised Learning to Biquality Learning: an Introduction
• Sustainable transparency in Recommender Systems: Bayesian Ranking of Images for Explainability
• DALLMi: Domain Adaption for LLM-based Multi-label Classifier
• Weakly Supervised AUC Optimization: A Unified Partial AUC Approach
• Positive unlabeled learning with tensor networks
• Towards Targeted Change Detection with Heterogeneous Remote Sensing Images for Forest Mortality Mapping
• Positive and unlabelled machine learning reveals new fast radio burst repeater candidates
• Optimised one-class classification performance
• Gaia Focused Product Release: A catalogue of sources around quasars to search for strongly lensed quasars
• Skill Disentanglement for Imitation Learning from Suboptimal Demonstrations
• Accurate Measures of Vaccination and Concerns of Vaccine Holdouts from Web Search Logs
• Importance Reweighting for Biquality Learning
• Zero-Shot Scene Graph Generation via Triplet Calibration and Reduction
• Noise-Resilient Point-wise Anomaly Detection in Time Series Using Weak Segment Labels
• Towards Improved Illicit Node Detection with Positive-Unlabelled Learning
• OOD-SEG: Exploiting out-of-distribution detection techniques for learning image segmentation from sparse multi-class positive-only annotations
• DomainHarvester: Harvesting Infrequently Visited Yet Trustworthy Domain Names
• Positive-Unlabelled Learning for identifying new candidate Dietary Restriction-related genes among Ageing-related genes
• Elemental Reactivity Maps for Materials Discovery
• Data-driven Discovery of Biophysical T Cell Receptor Co-specificity Rules
• Heterogeneous Domain Adaptation with Positive and Unlabeled Data
• Ca2+ transient detection and segmentation with the Astronomically motivated algorithm for Background Estimation And Transient Segmentation (Astro-BEATS)
• Automated all-sky detection of γ Doradus / δ Scuti hybrids in TESS data from positive unlabelled (PU) learning
• Critical Review for One-class Classification: recent advances and the reality behind them
• Feature-Weighted Maximum Representative Subsampling
• SAGE: Scalable Automatic Gating Ensemble for Confident Negative Harvesting in Fraud Detection
• Learning from Concealed Labels
• Positive-Unlabeled Constraint Learning for Inferring Nonlinear Continuous Constraints Functions from Expert Demonstrations
• ESA: Example Sieve Approach for Multi-Positive and Unlabeled Learning
• Perils of Label Indeterminacy: A Case Study on Prediction of Neurological Recovery After Cardiac Arrest
• DiVa: An Iterative Framework to Harvest More Diverse and Valid Labels from User Comments for Music
• Adaptive Gradient Calibration for Single-Positive Multi-Label Learning in Remote Sensing Image Scene Classification