Selective Labeling

Selective labeling focuses on strategically choosing which data points to label, maximizing the information gained from limited annotation resources. Current research emphasizes developing algorithms that mitigate biases introduced by selective labeling processes, often employing expectation-maximization frameworks, instrumental variable approaches, or curriculum learning within self-training models. This research is crucial for improving the efficiency and accuracy of machine learning models in various applications, particularly in domains with high annotation costs like medical image analysis and financial modeling, where obtaining fully labeled datasets is impractical. The resulting advancements promise to significantly reduce labeling expenses and improve model performance in data-scarce scenarios.

Papers

June 27, 2024

From Biased Selective Labels to Pseudo-Labels: An Expectation-Maximization Framework for Learning from Biased Decisions
Trenton Chang, Jenna Wiens
LeArning Abstract Label Bias Biased Decision Bias Label Selective Labeling

August 9, 2023

SLPT: Selective Labeling Meets Prompt Tuning on Label-Limited Lesion Segmentation
Fan Bai, Ke Yan, Xiaoyu Bai, Xinyu Mao, Xiaoli Yin, Jingren Zhou, Yu Shi, Le Lu, Max Q. -H. Meng
Prompt Tuning Style PROMPT Visual Prompt Tuning Prompt Based Pseudo Selective Labeling Label Constraint

June 13, 2023

Learning under Selective Labels with Data from Heterogeneous Decision-makers: An Instrumental Variable Approach
Jian Chen, Zhehao Li, Xiaojie Mao
LeArning Abstract Raw Data Decision Making Instrumental Variable Selection Bias Label Bias Heterogeneous Preference Synthetic Financial Selective Labeling

May 15, 2023

Algorithmic Censoring in Dynamic Learning Systems
Jennifer Chien, Margaret Roberts, Berk Ustun
Training Data Dynamic Learning Random Exploration False Negative Prediction Selective Labeling

March 30, 2023

C-SFDA: A Curriculum Learning Aided Self-Training Framework for Efficient Source Free Domain Adaptation
Nazmul Karim, Niluthpol Chowdhury Mithun, Abhinav Rajvanshi, Han-pang Chiu, Supun Samarasekera, Nazanin Rahnavard
Unsupervised Domain Adaptation Self Training Source Free Domain Adaptation Pseudo Label Refinement Selective Labeling

February 15, 2023

Deep Anomaly Detection under Labeling Budget Constraints
Aodong Li, Chen Qiu, Marius Kloft, Padhraic Smyth, Stephan Mandt, Maja Rudolph
Anomaly Detection Deep Anomaly Detection Semi Supervised Anomaly Detection Coverage Optimization Selective Labeling Label Budget

October 28, 2022

Radically Lower Data-Labeling Costs for Visually Rich Document Extraction Models
Yichao Zhou, James B. Wendt, Navneet Potti, Jing Xie, Sandeep Tata
Visually Rich Document Data Labeling Extraction Model Rich Document Extraction Task Selective Labeling

March 4, 2022

Selective Pseudo-labeling and Class-wise Discriminative Fusion for Sound Event Detection
Yunhao Liang, Yanhua Long, Yijie Li, Jiaen Liang
Feature Fusion Source Separation Separation Performance Sound Event Detection Sound Separation Audio Mixture Selective Labeling

Selective Labeling

Papers

From Biased Selective Labels to Pseudo-Labels: An Expectation-Maximization Framework for Learning from Biased Decisions

SLPT: Selective Labeling Meets Prompt Tuning on Label-Limited Lesion Segmentation

Learning under Selective Labels with Data from Heterogeneous Decision-makers: An Instrumental Variable Approach

Algorithmic Censoring in Dynamic Learning Systems

C-SFDA: A Curriculum Learning Aided Self-Training Framework for Efficient Source Free Domain Adaptation

Deep Anomaly Detection under Labeling Budget Constraints

Radically Lower Data-Labeling Costs for Visually Rich Document Extraction Models

Selective Pseudo-labeling and Class-wise Discriminative Fusion for Sound Event Detection