LaCViT: A Label-aware Contrastive Fine-tuning Framework for Vision Transformers
This addresses a critical challenge in enhancing generalization and transferability of ViTs for computer vision tasks, though it is incremental as it builds on existing fine-tuning methods.
The paper tackled the limitation of cross-entropy loss in Vision Transformers (ViTs) by introducing LaCViT, a label-aware contrastive fine-tuning framework, which improved performance by up to 10.78% in Top-1 Accuracy across eight image classification datasets.
Vision Transformers (ViTs) have emerged as popular models in computer vision, demonstrating state-of-the-art performance across various tasks. This success typically follows a two-stage strategy involving pre-training on large-scale datasets using self-supervised signals, such as masked random patches, followed by fine-tuning on task-specific labeled datasets with cross-entropy loss. However, this reliance on cross-entropy loss has been identified as a limiting factor in ViTs, affecting their generalization and transferability to downstream tasks. Addressing this critical challenge, we introduce a novel Label-aware Contrastive Training framework, LaCViT, which significantly enhances the quality of embeddings in ViTs. LaCViT not only addresses the limitations of cross-entropy loss but also facilitates more effective transfer learning across diverse image classification tasks. Our comprehensive experiments on eight standard image classification datasets reveal that LaCViT statistically significantly enhances the performance of three evaluated ViTs by up-to 10.78% under Top-1 Accuracy.