Belgian Health Data Agency (HDA)

> Github repository
In this paper, we explore the application of existing foundation models to cytological classification tasks, focusing on low-rank adaptation (LoRA), a parameter-efficient fine-tuning method well-suited to few-shot learning scenarios. We evaluate five foundation models across four cytological classification datasets. Our results demonstrate that fine-tuning the pre-trained backbones with LoRA significantly enhances model performance compared to merely fine-tuning the classifier head, achieving state-of-the-art results on both simple and complex classification tasks while requiring fewer data samples.