Balancing Accuracy, Calibration, and Efficiency in Active Learning with Vision Transformers Under Label Noise
This provides incremental guidance for practitioners deploying vision transformers in resource-constrained, noisy-label environments.
The study investigated how vision transformers perform under label noise in active learning, finding that larger ViT models like ViTl32 maintain better accuracy and calibration than smaller ones, while Swin Transformers are less robust, and active learning strategies only help at moderate noise levels.
Fine-tuning pre-trained convolutional neural networks on ImageNet for downstream tasks is well-established. Still, the impact of model size on the performance of vision transformers in similar scenarios, particularly under label noise, remains largely unexplored. Given the utility and versatility of transformer architectures, this study investigates their practicality under low-budget constraints and noisy labels. We explore how classification accuracy and calibration are affected by symmetric label noise in active learning settings, evaluating four vision transformer configurations (Base and Large with 16x16 and 32x32 patch sizes) and three Swin Transformer configurations (Tiny, Small, and Base) on CIFAR10 and CIFAR100 datasets, under varying label noise rates. Our findings show that larger ViT models (ViTl32 in particular) consistently outperform their smaller counterparts in both accuracy and calibration, even under moderate to high label noise, while Swin Transformers exhibit weaker robustness across all noise levels. We find that smaller patch sizes do not always lead to better performance, as ViTl16 performs consistently worse than ViTl32 while incurring a higher computational cost. We also find that information-based Active Learning strategies only provide meaningful accuracy improvements at moderate label noise rates, but they result in poorer calibration compared to models trained on randomly acquired labels, especially at high label noise rates. We hope these insights provide actionable guidance for practitioners looking to deploy vision transformers in resource-constrained environments, where balancing model complexity, label noise, and compute efficiency is critical in model fine-tuning or distillation.