Active Learning: Problem Settings and Recent Developments

Abstract

In supervised learning, acquiring labeled training data for a predictive model can be very costly, but acquiring a large amount of unlabeled data is often quite easy. Active learning is a method of obtaining predictive models with high precision at a limited cost through the adaptive selection of samples for labeling. This paper explains the basic problem settings of active learning and recent research trends. In particular, research on learning acquisition functions to select samples from the data for labeling, theoretical work on active learning algorithms, and stopping criteria for sequential data acquisition are highlighted. Application examples for material development and measurement are introduced.

arXiv comment31 pages, 2 figures

References in corpus (10)

• Auto-Encoding Variational Bayes
• Learning Active Learning from Data
• Two Optimal Strategies for Active Learning of Causal Models from Interventional Data
• Active One-shot Learning
• Efficient Construction Method for Phase Diagrams Using Uncertainty Sampling
• Stopping Active Learning based on Predicted Change of F Measure for Text Classification
• Analysis of Stopping Active Learning based on Stabilizing Predictions
• Stopping criterion for active learning based on deterministic generalization bounds
• Gradient Descent in RKHS with Importance Labeling
• Can Active Learning Experience Be Transferred?

Cited by in corpus (5)

• Graph Convolutional Networks based on Manifold Learning for Semi-Supervised Image Classification
• Topic-Guided Sampling For Data-Efficient Multi-Domain Stance Detection
• Stopping Criterion for Active Learning Based on Error Stability
• Impact of Stop Sets on Stopping Active Learning for Text Classification
• Improving Uncertainty Sampling with Bell Curve Weight Function