CLOSER: Towards Better Representation Learning for Few-Shot Class-Incremental Learning

TL;DR

This paper proposes a novel representation learning approach for few-shot class-incremental learning that emphasizes the importance of closer class features to improve transferability and discriminability, challenging traditional max-distance strategies.

Contribution

It introduces a counter-intuitive method focusing on feature spread within a confined space, backed by empirical and theoretical analysis, to enhance FSCIL performance.

Findings

Closer class features improve transferability and discriminability.

The proposed method outperforms traditional max-distance approaches.

The approach is validated through empirical results and information bottleneck theory.

Abstract

Aiming to incrementally learn new classes with only few samples while preserving the knowledge of base (old) classes, few-shot class-incremental learning (FSCIL) faces several challenges, such as overfitting and catastrophic forgetting. Such a challenging problem is often tackled by fixing a feature extractor trained on base classes to reduce the adverse effects of overfitting and forgetting. Under such formulation, our primary focus is representation learning on base classes to tackle the unique challenge of FSCIL: simultaneously achieving the transferability and the discriminability of the learned representation. Building upon the recent efforts for enhancing transferability, such as promoting the spread of features, we find that trying to secure the spread of features within a more confined feature space enables the learned representation to strike a better balance between transferability and discriminability. Thus, in stark contrast to prior beliefs that the inter-class distance should be maximized, we claim that the closer different classes are, the better for FSCIL. The empirical results and analysis from the perspective of information bottleneck theory justify our simple yet seemingly counter-intuitive representation learning method, raising research questions and suggesting alternative research directions. The code is available at https://github.com/JungHunOh/CLOSER_ECCV2024.