AI-based solutions in practice - from data management to deployment
The use of data by means of artificial intelligence (AI) creates many new possibilities. Important and necessary building blocks of all AI-based applications are the preparation and structuring of data. These are used to train powerful AI algorithms. This often requires different and also decentralized data sources, which must be bundled and organized in a uniform manner. For machine learning, moreover, it is not only the quantity of data that is important, but also its quality. Inaccurate or erroneous data can quickly lead to a significant reduction in the performance of AI-based applications. This can cause unnecessarily high costs and risks in many areas, such as control engineering, quality inspection or medical technology. Chimaera GmbH comes from the field of medical technology and has experience with the creation of AI-based medical products with high quality requirements for the data and the applications generated with it, which can be transferred very well to industrial projects in other fields.
In the talk, important aspects for the implementation of general AI-based projects will be discussed, starting with data management, annotation, up to the integration of customized AI solutions. Important aspects in planning and subsequent deployment will be highlighted and an attempt will be made to show where avoidable costs can arise in the implementation of AI projects. The emergence of an AI-based application is explained step-by-step, up to the planned, continuous further development involving different data sources by means of federated learning.
- Fundamentals of data management
- Creating high-quality annotations
- Cost-effective solution for data management and archiving
- Incorporating distributed data sources and federated learning
- Integration and deployment of AI-based solutions
Dr.-Ing. Marcus Prümmer studied computer engineering at the University of Mannheim. As a PhD student he held exercises on medical image processing at the Pattern Recognition Lab of the University of Erlangen-Nuremberg. Between 2005-2010 he was head of the medical image processing group, becoming senior researcher in 2009. In collaboration with Siemens Healthcare, he developed reconstruction algorithms for cardiac C-arm CT at Stanford University, Department of Radiology. He is founder and CEO of Chimaera GmbH since 2007, as well as responsible for the management of its AI department.
At a Glance
|Time:||17:00 - 18:30|
|Organization:||Bezirksverein Bayern Nordost e.V.|
NW Künstliche Intelligenz