Pu Learning

Positive-unlabeled (PU) learning tackles the challenge of training binary classifiers using only positively labeled and unlabeled data, a common scenario in many applications where obtaining negative labels is difficult or expensive. Current research focuses on improving the accuracy and robustness of PU learning algorithms, addressing issues like imbalanced data, noisy samples, and the violation of assumptions about how positive examples are selected for labeling. This involves developing novel loss functions, employing techniques like density estimation and pseudo-supervision, and adapting various model architectures, including neural networks and random forests. The advancements in PU learning have significant implications for diverse fields, such as medical image analysis, anomaly detection, and document classification, by enabling efficient and effective model training with limited labeled data.

Papers

July 9, 2024

PSPU: Enhanced Positive and Unlabeled Learning by Leveraging Pseudo Supervision
Chengjie Wang, Chengming Xu, Zhenye Gan, Jianlong Hu, Wenbing Zhu, Lizhuag Ma
Binary Classifier Industrial Anomaly Detection Positive Unlabeled Learning Positive Emotional Language Pseudo Supervision Unlabeled Learning Pu Learning

May 31, 2024

PUAL: A Classifier on Trifurcate Positive-Unlabeled Data
Xiaoke Wang, Xiaochen Yang, Rui Zhu, Jing-Hao Xue
Simple Classifier Positive Unlabeled Learning Bifurcation Diagram Pu Learning Positive Unlabeled Data Positive Unlabeled

May 7, 2024

Semi-Supervised Disease Classification based on Limited Medical Image Data
Yan Zhang, Chun Li, Zhaoxia Liu, Ming Li
Medical Image Semi Supervised Classification Pu Learning

May 3, 2024

Soft Label PU Learning
Puning Zhao, Jintao Deng, Xu Cheng
Pu Learning

January 20, 2024

Document Set Expansion with Positive-Unlabeled Learning: A Density Estimation-based Approach
Haiyang Zhang, Qiuyi Chen, Yuanjie Zou, Yushan Pan, Jia Wang, Mark Stevenson
Fine Grained Density Estimation Positive Unlabeled Learning Particle Density Estimation Method Pu Learning Document Expansion

December 27, 2023

Joint empirical risk minimization for instance-dependent positive-unlabeled data
Wojciech Rejchel, Paweł Teisseyre, Jan Mielniczuk
Empirical Risk Minimization Positive Unlabeled Learning Propensity Score Labeling Technique Pu Learning Conditional Probability

October 5, 2023

Learning A Disentangling Representation For PU Learning
Omar Zamzam, Haleh Akrami, Mahdi Soltanolkotabi, Richard Leahy
LeArning Abstract Loss Function Density Estimation Simulated Data Unsupervised Clustering Pu Learning

August 29, 2023

Class Prior-Free Positive-Unlabeled Learning with Taylor Variational Loss for Hyperspectral Remote Sensing Imagery
Hengwei Zhao, Xinyu Wang, Jingtao Li, Yanfei Zhong
LeArning Abstract Natural Gradient Hyperspectral Image Classification Class Prior Pu Learning Hyperspectral Remote Sensing Variational Loss

March 21, 2023

Dens-PU: PU Learning with Density-Based Positive Labeled Augmentation
Vasileios Sevetlidis, George Pavlidis, Spyridon Mouroutsos, Antonios Gasteratos
Anomaly Detection Binary Classification Latent Encoding Label Augmentation Pu Learning Positive Unlabeled Data

March 14, 2023

Positive Unlabeled Learning Selected Not At Random (PULSNAR): class proportion estimation when the SCAR assumption does not hold
Praveen Kumar, Christophe G. Lambert
Positive Unlabeled Learning Random Participation Semi Supervised Classification Pu Learning Class Distribution Estimation Positive Example Scar Assumption

March 8, 2023

Automatic Debiased Learning from Positive, Unlabeled, and Exposure Data
Masahiro Kato, Shuting Wu, Kodai Kureishi, Shota Yasui
Binary Classification Selection Bias Debiasing Framework Positive Emotional Language Model Debiasing Pu Learning Positive Label Exposure Data

January 25, 2023

PULL: Reactive Log Anomaly Detection Based On Iterative PU Learning
Thorsten Wittkopp, Dominik Scheinert, Philipp Wiesner, Alexander Acker, Odej Kao
Anomaly Detection Log Anomaly Detection Pu Learning Push Pull

October 16, 2022

Positive-Unlabeled Learning using Random Forests via Recursive Greedy Risk Minimization
Jonathan Wilton, Abigail M. Y. Koay, Ryan K. L. Ko, Miao Xu, Nan Ye
Random Forest Positive Unlabeled Learning Greedy Learning Pu Learning