Boosting Few-Shot Classification with View-Learnable Contrastive Learning

TL;DR

This paper introduces a view-learnable contrastive learning approach for few-shot classification, enhancing the discrimination of fine-grained subcategories by automatically generating more accurate image views, leading to improved performance.

Contribution

It proposes a novel contrastive loss integrated with a learn-to-learn view generation method for better fine-grained discrimination in few-shot learning.

Findings

Outperforms existing few-shot classification methods on standard benchmarks.

Effectively learns latent fine-grained structures in embedding space.

Automatically generates more accurate views, reducing noise in contrastive learning.

Abstract

The goal of few-shot classification is to classify new categories with few labeled examples within each class. Nowadays, the excellent performance in handling few-shot classification problems is shown by metric-based meta-learning methods. However, it is very hard for previous methods to discriminate the fine-grained sub-categories in the embedding space without fine-grained labels. This may lead to unsatisfactory generalization to fine-grained subcategories, and thus affects model interpretation. To tackle this problem, we introduce the contrastive loss into few-shot classification for learning latent fine-grained structure in the embedding space. Furthermore, to overcome the drawbacks of random image transformation used in current contrastive learning in producing noisy and inaccurate image pairs (i.e., views), we develop a learning-to-learn algorithm to automatically generate different views of the same image. Extensive experiments on standard few-shot learning benchmarks demonstrate the superiority of our method.