Deep Transfer Learning (Glossary)

The term Deep Transfer Learning belongs in the categories Artificial Intelligence, Digital Transformation, and Industry and Industry 4.0.

Deep Transfer Learning describes a modern method of machine learning. In this approach, an artificial intelligence (AI) first learns a task, such as image recognition, using large amounts of data. This knowledge is then transferred to a new, but similar task, allowing the AI to learn there much faster and with less data.

A simple example: An AI is trained to recognise cats in photos. Afterwards, this AI can use its „knowledge“ to identify dogs in photos. The basic techniques of image recognition – such as recognising fur or ears – remain useful, even if the tasks are different. This saves a company a lot of time and expensive data.

Deep transfer learning thus simplifies the introduction of artificial intelligence into new business areas and production processes. Companies benefit from this because existing models are reused and the effort required for new AI applications is significantly reduced. This makes the technology particularly interesting for companies that want to react quickly and flexibly to trends.