mixmatch
None
view repo
Semi-supervised learning has proven to be a powerful paradigm for leveraging unlabeled data to mitigate the reliance on large labeled datasets. In this work, we unify the current dominant approaches for semi-supervised learning to produce a new algorithm, MixMatch, that works by guessing low-entropy labels for data-augmented unlabeled examples and mixing labeled and unlabeled data using MixUp. We show that MixMatch obtains state-of-the-art results by a large margin across many datasets and labeled data amounts. For example, on CIFAR-10 with 250 labels, we reduce error rate by a factor of 4 (from 38 a factor of 2 on STL-10. We also demonstrate how MixMatch can help achieve a dramatically better accuracy-privacy trade-off for differential privacy. Finally, we perform an ablation study to tease apart which components of MixMatch are most important for its success.
READ FULL TEXTNone
Code for "MixMatch - A Holistic Approach to Semi-Supervised Learning"
Pytorch Implementation of the paper MixMatch: A Holistic Approach to Semi-Supervised Learning (https://arxiv.org/pdf/1905.02249.pdf)
Reimplementation of "Realistic Evaluation of Deep Semi-Supervised Learning Algorithms"
Code for the paper: RealMix: Towards Realistic Semi-Supervised Deep Learning Algorithms