Fraunhofer Institute for Intelligent Analysis and Information Systems IAISMachine Learning (ML) is a key technology in relation to cognitive systems and Artificial Intelligence (AI). Thanks to progressively cheaper and more powerful sensors and processors ML techniques are increasingly driving digitalization and are determining factors in maintaining a competitive edge in many sectors. There is hardly any sector that isn’t currently undergoing a substantial transformation because of ML- and AI-based technology: from manufacturing and logistics through to medicine.
We have been at the forefront of applied ML research for many years. We shape the future of Machine Learning: modular Machine Learning, driven by data and knowledge. Our researchers are developing people-centered, comprehensible and reliable solutions.
We are also working on resource efficient applications that can be run on the most diverse range of hardware platforms. Our processes are tailored to specific application scenarios, from distributed Learning (Edge Computing) to quantum Machine Learning. Our learning algorithms can now be used for applications that have previously not been thought possible for commerce or society.
We are reporting on this technology, including its potentials and limitations in studies and scientific publications. We are actively involved in central research projects in Germany and the rest of Europe and keep a keen eye on the social debate over the impact intelligent machines might have on our lives.
We design the Machine Learning of the future, which is driven by both data and knowledge.
© sakkmesterke / fotolia.de
Machine Learning
Research priorities
Big Data Analytics
Machine Learning involving extremely large volumes of data which are processed in real time on scalable platforms and infrastructures
Decentralized Machine Learning (Edge ML)
Learning on user devices or in the cloud without transmitting sensitive data in order to shorten response times and increase the efficient usage of distributed data
Informed Machine Learning
A combination of statistical learning and knowledge-driven approaches such as semantic methods based on knowledge graphs or physical system simulations
Text Analytics
The process used to analyze different kinds of texts in order to extract and summarize information or recognizing moods
Quantum Learning
The conversion of learning algorithms onto quantum computers for the purpose of solving computation-intensive problems that current, conventional computers are unable to solve
Visual and interactive Analytics
Processes aimed at performing explorative, interactive data analyses based on the visual abilities of people; main focuses are mobility prognoses and interactive model making