Support Set
A "support set" in machine learning refers to a small, carefully selected subset of labeled data used to improve model performance in scenarios with limited training data, such as few-shot learning and domain adaptation. Current research focuses on optimizing support set selection methods, including algorithms that leverage self-supervision, clustering techniques, and information-theoretic metrics to identify the most informative examples. These advancements aim to enhance model generalization and efficiency, particularly in applications like image recognition, video captioning, and natural language processing, where obtaining large labeled datasets can be costly or impractical.
Papers
November 7, 2024
August 14, 2024
May 24, 2024
April 25, 2024
August 9, 2023
February 27, 2023
December 9, 2022
May 19, 2022