Embedding Space Allocation with Angle-Norm Joint Classifiers for Few-Shot Class-Incremental Learning

TL;DR

This paper introduces SAAN, a novel framework for few-shot class-incremental learning that allocates embedding space effectively and uses angle-norm joint classifiers to improve learning of new classes from limited data.

Contribution

SAAN provides a new space allocation strategy and angle-norm joint classifiers, addressing class imbalance and space utilization issues in FSCIL, and can enhance existing methods.

Findings

Achieves state-of-the-art performance on FSCIL benchmarks.

Effectively allocates feature space to prevent class interference.

Improves classification accuracy with angle-norm joint logits.

Abstract

Few-shot class-incremental learning (FSCIL) aims to continually learn new classes from only a few samples without forgetting previous ones, requiring intelligent agents to adapt to dynamic environments. FSCIL combines the characteristics and challenges of class-incremental learning and few-shot learning: (i) Current classes occupy the entire feature space, which is detrimental to learning new classes. (ii) The small number of samples in incremental rounds is insufficient for fully training. In existing mainstream virtual class methods, for addressing the challenge (i), they attempt to use virtual classes as placeholders. However, new classes may not necessarily align with the virtual classes. For the challenge (ii), they replace trainable fully connected layers with Nearest Class Mean (NCM) classifiers based on cosine similarity, but NCM classifiers do not account for sample imbalance issues. To address these issues in previous methods, we propose the class-center guided embedding Space Allocation with Angle-Norm joint classifiers (SAAN) learning framework, which provides balanced space for all classes and leverages norm differences caused by sample imbalance to enhance classification criteria. Specifically, for challenge (i), SAAN divides the feature space into multiple subspaces and allocates a dedicated subspace for each session by guiding samples with the pre-set category centers. For challenge (ii), SAAN establishes a norm distribution for each class and generates angle-norm joint logits. Experiments demonstrate that SAAN can achieve state-of-the-art performance and it can be directly embedded into other SOTA methods as a plug-in, further enhancing their performance.