Flatness Improves Backbone Generalisation in Few-shot Classification

arXiv:2404.07696v1 Announce Type: new
Abstract: Deployment of deep neural networks in real-world settings typically requires adaptation to new tasks with few examples. Few-shot classification (FSC) provides a solution to this problem by leveraging pre-trained backbones for fast adaptation to new classes. Surprisingly, most efforts have only focused on developing architectures for easing the adaptation to the target domain without considering the importance of backbone training for good generalisation. We show that flatness-aware backbone training with vanilla fine-tuning results in a …

abstract architectures arxiv classification cs.cv cs.lg deployment examples few-shot networks neural networks solution tasks type world