Importance Weighted Active Learning

Abstract

We present a practical and statistically consistent scheme for actively learning binary classifiers under general loss functions. Our algorithm uses importance weighting to correct sampling bias, and by controlling the variance, we are able to give rigorous label complexity bounds for the learning process. Experiments on passively labeled data show that this approach reduces the label complexity required to achieve good predictive performance on many learning problems.

Cited by in corpus (9)

• Active Learning for Graph Neural Networks via Node Feature Propagation
• The Unreasonable Effectiveness of Noisy Data for Fine-Grained Recognition
• Reinforcement Learning and Adaptive Sampling for Optimized DNN Compilation
• Online Active Model Selection for Pre-trained Classifiers
• Loss-Proportional Subsampling for Subsequent ERM
• Efficient Diverse Ensemble for Discriminative Co-Tracking
• Counterfactual Evaluation of Treatment Assignment Functions with Networked Observational Data
• Active Learning under Label Shift
• The Utility of Abstaining in Binary Classification