A Survey on Few-Shot Class-Incremental Learning

TL;DR

This survey comprehensively reviews the field of few-shot class-incremental learning (FSCIL), synthesizing theoretical and applied research, categorizing methods, evaluating performance, and discussing future directions to enhance model adaptability with limited data.

Contribution

The paper introduces a novel categorization of FSCIL methods, reviews over 50 studies, and evaluates recent research on benchmark datasets, providing a comprehensive analysis from multiple perspectives.

Findings

FSCIL methods are categorized into five subtypes.

Recent research shows promising results on benchmark datasets.

FSCIL has achieved significant progress in computer vision and NLP applications.

Abstract

Large deep learning models are impressive, but they struggle when real-time data is not available. Few-shot class-incremental learning (FSCIL) poses a significant challenge for deep neural networks to learn new tasks from just a few labeled samples without forgetting the previously learned ones. This setup easily leads to catastrophic forgetting and overfitting problems, severely affecting model performance. Studying FSCIL helps overcome deep learning model limitations on data volume and acquisition time, while improving practicality and adaptability of machine learning models. This paper provides a comprehensive survey on FSCIL. Unlike previous surveys, we aim to synthesize few-shot learning and incremental learning, focusing on introducing FSCIL from two perspectives, while reviewing over 30 theoretical research studies and more than 20 applied research studies. From the theoretical perspective, we provide a novel categorization approach that divides the field into five subcategories, including traditional machine learning methods, meta-learning based methods, feature and feature space-based methods, replay-based methods, and dynamic network structure-based methods. We also evaluate the performance of recent theoretical research on benchmark datasets of FSCIL. From the application perspective, FSCIL has achieved impressive achievements in various fields of computer vision such as image classification, object detection, and image segmentation, as well as in natural language processing and graph. We summarize the important applications. Finally, we point out potential future research directions, including applications, problem setups, and theory development. Overall, this paper offers a comprehensive analysis of the latest advances in FSCIL from a methodological, performance, and application perspective.