BoSS: A Best-of-Strategies Selector as an Oracle for Deep Active Learning

Researchers introduce BoSS (Best-of-Strategies Selector), a new oracle strategy for active learning that outperforms existing methods by using an ensemble approach to select optimal data annotation batches. The study reveals that current state-of-the-art active learning strategies still significantly underperform compared to oracle performance, particularly on large-scale datasets.

• →BoSS introduces a scalable oracle strategy that constructs candidate batches through ensemble selection methods for improved active learning performance.
• →Current state-of-the-art active learning strategies show notable performance gaps compared to oracle strategies, especially on large datasets with many classes.
• →The ensemble-based approach demonstrates superior scalability for large-scale active learning scenarios with complex deep neural networks.
• →BoSS can be easily extended with new strategies as they emerge, ensuring future adaptability as a reliable oracle method.
• →Ensemble-based selection approaches may provide a solution to address inconsistent performance issues in existing active learning strategies.