Read the original document by opening this link in a new tab.
Table of Contents
1 Introduction
2 Related Work
2.1 Active Learning
2.2 Unsupervised Representation Learning
3 Methodology
3.1 O -the-Shelf Features for Active Learning
3.2 Knowledge Clusters inside Images
3.3 Sample Selection Strategy
4 Results
5 Conclusion
Summary
The paper discusses a novel general and efficient active learning method called GEAL, which challenges the traditional pipeline of active learning. It emphasizes the importance of utilizing knowledge clusters extracted from intermediate features for selecting data samples in one-shot without the need for repetitive model training. The method aims to improve efficiency by reducing the time complexity significantly compared to existing approaches. Extensive experiments validate the effectiveness and efficiency of the proposed method across various tasks and datasets.