Calibrated Uncertainty Sampling for Active Learning
This addresses the issue of sub-optimal generalization and high calibration error in active learning, particularly for deep neural networks, which is an incremental improvement for machine learning practitioners.
The paper tackles the problem of active learning for classifiers with low calibration error by proposing a new acquisition function that estimates calibration errors before using model uncertainty, leading to bounded calibration error on unlabeled and test data. Empirically, it achieves lower calibration and generalization errors compared to baselines in pool-based active learning settings.
We study the problem of actively learning a classifier with a low calibration error. One of the most popular Acquisition Functions (AFs) in pool-based Active Learning (AL) is querying by the model's uncertainty. However, we recognize that an uncalibrated uncertainty model on the unlabeled pool may significantly affect the AF effectiveness, leading to sub-optimal generalization and high calibration error on unseen data. Deep Neural Networks (DNNs) make it even worse as the model uncertainty from DNN is usually uncalibrated. Therefore, we propose a new AF by estimating calibration errors and query samples with the highest calibration error before leveraging DNN uncertainty. Specifically, we utilize a kernel calibration error estimator under the covariate shift and formally show that AL with this AF eventually leads to a bounded calibration error on the unlabeled pool and unseen test data. Empirically, our proposed method surpasses other AF baselines by having a lower calibration and generalization error across pool-based AL settings.