KNN Transformer with Pyramid Prompts for Few-Shot Learning
Few-Shot Learning (FSL) aims to recognize new classes with limited labeled data. Recent
studies have attempted to address the challenge of rare samples with textual prompts to …
studies have attempted to address the challenge of rare samples with textual prompts to …
Enhancing Few-shot Classification through Token Selection for Balanced Learning
W Zeng, P Sun, H Zhang - 2024 International Joint Conference …, 2024 - ieeexplore.ieee.org
In recent years, patch-based approaches have shown promise in few-shot learning, with
further improvements observed through the use of self-supervised learning. However, we …
further improvements observed through the use of self-supervised learning. However, we …