Bridging Diversity and Uncertainty in Active learning with Self-Supervised Pre-Training
This work addresses the cold start problem in active learning for researchers, but it is incremental as it combines existing strategies with a heuristic approach.
This study tackled the challenge of integrating diversity-based and uncertainty-based sampling in active learning by introducing the TCM heuristic, which uses TypiClust for diversity and Margin for uncertainty, and demonstrated that it consistently outperforms existing methods across datasets in low and high data regimes.
This study addresses the integration of diversity-based and uncertainty-based sampling strategies in active learning, particularly within the context of self-supervised pre-trained models. We introduce a straightforward heuristic called TCM that mitigates the cold start problem while maintaining strong performance across various data levels. By initially applying TypiClust for diversity sampling and subsequently transitioning to uncertainty sampling with Margin, our approach effectively combines the strengths of both strategies. Our experiments demonstrate that TCM consistently outperforms existing methods across various datasets in both low and high data regimes.