self-supervised learning (SSL)

Self-supervised learning (SSL) is a way for machines to learn by generating labels from the input data instead of using labels from people. This approach is particularly useful when dealing with large amounts of unlabeled data, which is common in many real-world applications.

With self-supervised learning, the model creates its own labels by performing a pretext task, helping it learn useful data representations without getting labels directly from humans. For example, in natural language processing (NLP), a model might predict the next word in a sentence given the previous words, or in computer vision, it might predict the color version of a grayscale image.

Self-supervised learning is distinct from unsupervised learning, which also deals with unlabeled data but does not involve the creation of pretext tasks. Instead, unsupervised learning methods focus on clustering or dimensionality reduction. Self-supervised learning is also different from supervised learning, which requires a large amount of labeled data, and semi-supervised learning, which uses a combination of labeled and unlabeled data.

For more information:

1. https://en.wikipedia.org/wiki/Self-supervised_learning
2. https://neptune.ai/blog/self-supervised-learning
3. https://encord.com/blog/self-supervised-learning/
4. https://www.v7labs.com/blog/self-supervised-learning-guide
5. https://www.exforsys.com/career-center/self-supervision/self-supervision-introduction.html
6. https://viso.ai/deep-learning/self-supervised-learning-for-computer-vision/
7. https://www.linkedin.com/pulse/meaning-self-supervision-wade-townsend-ii