Robust Contrastive Active Learning with Feature-guided Query Strategies. (arXiv:2109.06873v2 [cs.LG] UPDATED)

We introduce supervised contrastive active learning (SCAL) and propose
efficient query strategies in active learning based on the feature similarity
(featuresim) and principal component analysis based feature-reconstruction
error (fre) to select informative data samples with diverse feature
representations. We demonstrate our proposed method achieves state-of-the-art
accuracy, model calibration and reduces sampling bias in an active learning
setup for balanced and imbalanced datasets on image classification tasks. We
also evaluate robustness of model to distributional shift derived from
different query strategies …

active learning arxiv feature learning lg query strategies