Towards Computationally Feasible Deep Active Learning
This work addresses computational feasibility issues in deep active learning for text classification and tagging, offering incremental improvements to reduce resource requirements.
The paper tackles the computational burden of deep active learning by proposing two techniques that reduce iteration duration and overhead, and demonstrates that their algorithm overcomes the performance gap between acquisition and successor models, achieving higher performance with a smaller, faster acquisition model.
Active learning (AL) is a prominent technique for reducing the annotation effort required for training machine learning models. Deep learning offers a solution for several essential obstacles to deploying AL in practice but introduces many others. One of such problems is the excessive computational resources required to train an acquisition model and estimate its uncertainty on instances in the unlabeled pool. We propose two techniques that tackle this issue for text classification and tagging tasks, offering a substantial reduction of AL iteration duration and the computational overhead introduced by deep acquisition models in AL. We also demonstrate that our algorithm that leverages pseudo-labeling and distilled models overcomes one of the essential obstacles revealed previously in the literature. Namely, it was shown that due to differences between an acquisition model used to select instances during AL and a successor model trained on the labeled data, the benefits of AL can diminish. We show that our algorithm, despite using a smaller and faster acquisition model, is capable of training a more expressive successor model with higher performance.