Self-Supervised Learning is a term from the fields of Artificial Intelligence, Big Data and Smart Data, and Automation. It refers to an innovative method by which computers can learn independently from large amounts of data, without the need for extensive manual data preparation or labelling by humans.
In self-supervised learning, machines attempt to extract existing information from unstructured data and recognise patterns – similar to how humans often learn new things through observation. What's special is that the computer sets itself tasks, which it then tries to solve using the available data.
A simple example: Imagine a computer is meant to understand how sentences are constructed. To do this, it's given many texts, some of which have a word omitted. Its task: to guess the missing word. Through many such puzzles, the computer recognises relationships in the texts without a human having to comment on each sentence individually.
Self-supervised learning makes it easier for companies to use their own data and make automated systems smarter – from voice assistants to automated quality controls in industry. This enables powerful AI applications to be created, even when no expensively labelled training data is available.
Deutsch 
English (UK)