Zhang Jinghua, Liu Li, Silven Olli, Pietikainen Matti, Hu Dewen
IEEE Trans Pattern Anal Mach Intell. 2025 Apr;47(4):2924-2945. doi: 10.1109/TPAMI.2025.3529038. Epub 2025 Mar 6.
Few-shot Class-Incremental Learning (FSCIL) presents a unique challenge in Machine Learning (ML), as it necessitates the Incremental Learning (IL) of new classes from sparsely labeled training samples without forgetting previous knowledge. While this field has seen recent progress, it remains an active exploration area. This paper aims to provide a comprehensive and systematic review of FSCIL. In our in-depth examination, we delve into various facets of FSCIL, encompassing the problem definition, the discussion of the primary challenges of unreliable empirical risk minimization and the stability-plasticity dilemma, general schemes, and relevant problems of IL and Few-shot Learning (FSL). Besides, we offer an overview of benchmark datasets and evaluation metrics. Furthermore, we introduce the Few-shot Class-incremental Classification (FSCIC) methods from data-based, structure-based, and optimization-based approaches and the Few-shot Class-incremental Object Detection (FSCIOD) methods from anchor-free and anchor-based approaches. Beyond these, we present several promising research directions within FSCIL that merit further investigation.
少样本类别增量学习（FSCIL）在机器学习（ML）中提出了一个独特的挑战，因为它需要从稀疏标注的训练样本中对新类别进行增量学习（IL），同时又不能忘记先前的知识。尽管该领域最近取得了进展，但它仍然是一个活跃的探索领域。本文旨在对FSCIL进行全面而系统的综述。在我们的深入研究中，我们深入探讨了FSCIL的各个方面，包括问题定义、对不可靠经验风险最小化和稳定性-可塑性困境等主要挑战的讨论、一般方案以及IL和少样本学习（FSL）的相关问题。此外，我们概述了基准数据集和评估指标。此外，我们介绍了基于数据、基于结构和基于优化的方法的少样本类别增量分类（FSCIC）方法，以及基于无锚点和基于锚点的方法的少样本类别增量目标检测（FSCIOD）方法。除此之外，我们还提出了FSCIL中几个值得进一步研究的有前景的研究方向。
