FALCON: False-Negative Aware Learning of Contrastive Negatives in Vision-Language Alignment
This addresses a critical challenge in vision-language alignment for AI researchers, though it appears incremental as an enhancement to existing frameworks.
The paper tackles the problem of false negatives degrading vision-language pretraining by proposing FALCON, a learning-based mini-batch construction strategy that adaptively balances hard and false negatives. Results show it significantly improves performance across three frameworks and various downstream tasks.
False negatives pose a critical challenge in vision-language pretraining (VLP) due to the many-to-many correspondence between images and texts in large-scale datasets. These false negatives introduce conflicting supervision signals that degrade the learned embedding space and diminish the effectiveness of hard negative sampling. In this paper, we propose FALCON (False-negative Aware Learning of COntrastive Negatives), a learning-based mini-batch construction strategy that adaptively balances the trade-off between hard and false negatives during VLP. Rather than relying on fixed heuristics, FALCON employs a negative mining scheduler that dynamically selects negative samples of appropriate hardness for each anchor instance during mini-batch construction, guided by a proxy for cross-modal alignment improvement. Experimental results demonstrate that FALCON significantly improves performance across three vision-language learning frameworks (ALBEF, BLIP-2, SigLIP-2) and a broad range of downstream tasks and evaluation settings, underscoring its effectiveness and robustness in mitigating the impact of false negatives.