Deep learning is the area of ​​artificial intelligence (AI) in which great progress has been made in recent years. Most new methods in this area are developed in the context of image and video data from everyday scenes. A transfer to the data and tasks available in the health industry, pharmaceutical, chemical and food production is promising, but not possible without special research. M²Aind-DeepLearning rises to this challenge in order to make use of the advantages of the high innovation potential of deep learning methods for other fields of application in the impulse partnership. While deep learning is part of all M²Aind impulse projects, the concrete work packages of this project each pursue additional topic-centric solutions. The topics worked on are selected in such a way that the solutions examined have application potential in various sectors of the M²Aind impulse partnership and offer effective interfaces with other impulse projects.

In M²Aind-DeepLearning, innovations in the three additional application fields of life cycle monitoring, imaging mass spectrometry and hygiene monitoring are to be researched on a fundamental basis. In the area of ​​life cycle monitoring, methods for modeling or generating artificial domain-specific data for the hybrid training of deep learning models are being investigated. In the field of mass spectrometry, an interactive web-based platform for deep learning experiments is being created. In the field of hygiene monitoring, methods for the detection of microorganisms in Petri dishes are to be researched. The challenges of all work packages lie in particular in the development of suitable methods and in the preparation of suitable learning data.