RankCLIP: Ranking-Consistent Language-Image Pretraining
This addresses the limitation of existing models in handling complex multimodal relationships, offering an incremental improvement for vision-language tasks.
The paper tackles the problem of rigid one-to-one mappings in vision-language models like CLIP by introducing RankCLIP, a pre-training method that uses list-wise loss and ranking consistency to capture many-to-many relationships, resulting in significant gains in zero-shot classification over state-of-the-art methods.
Self-supervised contrastive learning models, such as CLIP, have set new benchmarks for vision-language models in many downstream tasks. However, their dependency on rigid one-to-one mappings overlooks the complex and often multifaceted relationships between and within texts and images. To this end, we introduce RankCLIP, a novel pre-training method that extends beyond the rigid one-to-one matching framework of CLIP and its variants. By extending the traditional pair-wise loss to list-wise, and leveraging both in-modal and cross-modal ranking consistency, RankCLIP improves the alignment process, enabling it to capture the nuanced many-to-many relationships between and within each modality. Through comprehensive experiments, we demonstrate the effectiveness of RankCLIP in various downstream tasks, notably achieving significant gains in zero-shot classifications over state-of-the-art methods, underscoring the importance of this enhanced learning process.