Reproduce “An Overview of Deep Semi-Supervised Learning” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yassine Ouali, Céline Hudelot, Myriam Tami

A comprehensive overview of deep semi-supervised learning (SSL) systematically categorizes and details dominant approaches that leverage limited labeled data alongside abundant unlabeled data for deep neural networks. The survey covers methods across consistency regularization, proxy-labeling, generative models, and graph-based techniques, outlining best practices for evaluation.

Key insights

  • Combining multiple semi-supervised learning principles into "holistic methods" often yields superior performance.
  • Self-supervised learning effectively provides rich, transferable features from unlabeled data, enhancing deep semi-supervised learning.
  • Rigorous and standardized evaluation methodologies are crucial for comparable and reproducible research in deep semi-supervised learning.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.