self-supervised learning
A family of techniques for converting an unsupervised machine learning problem into a supervised machine learning problem by creating surrogate labels from unlabeled examples.
Plain English Explanation
A family of techniques for converting an unsupervised machine learning problem into a supervised machine learning problem by creating surrogate labels from unlabeled examples. Some Transformer-based models such as BERT use self-supervised learning. Self-supervised training is a semi-supervised learning approach.
How is it used?
Practitioners refer to self-supervised learning when building, training, or evaluating machine learning systems. It appears in research papers, product documentation, and technical discussions about AI capabilities and limitations.