Active Learning
Active learning is a machine learning paradigm focused on optimizing data labeling efficiency by strategically selecting the most informative samples for annotation from a larger unlabeled pool. Current research emphasizes developing novel acquisition functions and data pruning strategies to reduce computational costs associated with large datasets, exploring the integration of active learning with various model architectures (including deep neural networks, Gaussian processes, and language models), and addressing challenges like privacy preservation and handling open-set noise. This approach holds significant promise for reducing the substantial cost and effort of data labeling in diverse fields, ranging from image classification and natural language processing to materials science and healthcare.
Papers
Modular Deep Active Learning Framework for Image Annotation: A Technical Report for the Ophthalmo-AI Project
Md Abdul Kadir, Hasan Md Tusfiqur Alam, Pascale Maul, Hans-Jürgen Profitlich, Moritz Wolf, Daniel Sonntag
Active Learning for Regression based on Wasserstein distance and GroupSort Neural Networks
Benjamin Bobbia, Matthias Picard
Generative Active Learning for Image Synthesis Personalization
Xulu Zhang, Wengyu Zhang, Xiao-Yong Wei, Jinlin Wu, Zhaoxiang Zhang, Zhen Lei, Qing Li
Improve Cost Efficiency of Active Learning over Noisy Dataset
Zan-Kai Chong, Hiroyuki Ohsaki, Bryan Ng
STAR: Constraint LoRA with Dynamic Active Learning for Data-Efficient Fine-Tuning of Large Language Models
Linhai Zhang, Jialong Wu, Deyu Zhou, Guoqiang Xu
Feature Alignment: Rethinking Efficient Active Learning via Proxy in the Context of Pre-trained Models
Ziting Wen, Oscar Pizarro, Stefan Williams