BriefGPT.xyz
Aug, 2023
通过连接高和低置信度预测增强半监督学习
Boosting Semi-Supervised Learning by bridging high and low-confidence predictions
HTML
PDF
Khanh-Binh Nguyen, Joon-Sung Yang
TL;DR
提出了一种名为ReFixMatch的新方法,旨在利用所有未标记数据进行训练,从而提高模型的泛化能力和在半监督学习基准测试上的性能。值得注意的是,ReFixMatch在ImageNet上使用10万个标记示例时达到了41.05%的top-1准确率,优于基准FixMatch和目前最先进的方法。
Abstract
pseudo-labeling
is a crucial technique in
semi-supervised learning
(SSL), where artificial labels are generated for unlabeled data by a trained model, allowing for the simultaneous training of labeled and unlabel
→