So sánh phương pháp

	Học bán giám sát có điều chuẩn	Học tăng cường tự giám sát
Lĩnh vực	Học máy	Học máy
Họ	Machine learning	Machine learning
Năm ra đời	2006	2018–2020
Người khởi xướng	Belkin, M.; Niyogi, P.; Sindhwani, V.	LeCun, Y. and community (formalized ~2018–2020)
Loại	Regularized learning paradigm	Representation learning paradigm
Công trình gốc	Belkin, M., Niyogi, P., & Sindhwani, V. (2006). Manifold regularization: A geometric framework for learning from labeled and unlabeled examples. Journal of Machine Learning Research, 7, 2399–2434.	LeCun, Y. & Misra, I. (2022). Self-supervised learning: The dark matter of intelligence. Meta AI Blog. https://ai.facebook.com/blog/self-supervised-learning-the-dark-matter-of-intelligence/
Tên gọi khác	manifold regularization, graph-regularized SSL, semi-supervised regularization, Laplacian regularization	SSL, self-supervised pre-training, pretext-task learning, unsupervised representation learning
Liên quan	6	3
Tóm tắt	Regularized semi-supervised learning adds explicit geometric or graph-based penalty terms to a semi-supervised objective so that the decision function varies smoothly over the data manifold. Pioneered through manifold regularization (Belkin, Niyogi & Sindhwani, 2006), it exploits the structure of both labeled and unlabeled examples to learn more accurate models than supervised regularization alone when labeled data are scarce.	Self-supervised learning (SSL) is a machine-learning paradigm that generates its own supervisory signal directly from unlabeled data by defining an auxiliary pretext task — such as predicting masked words, rotating images, or contrasting augmented views — and uses the learned representations as a powerful starting point for downstream tasks with minimal labeled examples.