Coarse-To-Fine Incremental Few-Shot Learning
This addresses the challenge of incremental few-shot recognition for AI systems that need to adapt to new, detailed classes without catastrophic forgetting, though it is an incremental improvement over existing CIL/FSCIL methods.
The paper tackles the problem of class-incremental learning where models must recognize novel, finer-grained classes over time without forgetting pre-trained coarse classes, proposing a strategy called Knowe that learns embeddings from coarse labels and freezes classifiers from fine labels, achieving state-of-the-art performance on datasets like CIFAR-100, BREEDS, and tieredImageNet.
Different from fine-tuning models pre-trained on a large-scale dataset of preset classes, class-incremental learning (CIL) aims to recognize novel classes over time without forgetting pre-trained classes. However, a given model will be challenged by test images with finer-grained classes, e.g., a basenji is at most recognized as a dog. Such images form a new training set (i.e., support set) so that the incremental model is hoped to recognize a basenji (i.e., query) as a basenji next time. This paper formulates such a hybrid natural problem of coarse-to-fine few-shot (C2FS) recognition as a CIL problem named C2FSCIL, and proposes a simple, effective, and theoretically-sound strategy Knowe: to learn, normalize, and freeze a classifier's weights from fine labels, once learning an embedding space contrastively from coarse labels. Besides, as CIL aims at a stability-plasticity balance, new overall performance metrics are proposed. In that sense, on CIFAR-100, BREEDS, and tieredImageNet, Knowe outperforms all recent relevant CIL/FSCIL methods that are tailored to the new problem setting for the first time.