Applied AI and Big Data

Many application areas increasingly rely on data analytics, machine learning, and AI methods. It is therefore an important goal of ScaDS.AI Dresden/Leipzig to develop applied AI and Big Data in Life Sciences and Medicine, Environmental and Earth Sciences, Software Engineering, Physics and Chemistry, as well as Engineering and Business.

Core topics of the applied AI and Big Data research area include data-driven modeling, surrogate models, physics-informed machine learning, and inverse design using AI and Machine Learning approaches suitably adapted to the challenges of the specific application domain. Data Science and Machine Learning methods like sparse multidimensional inference, physics-informed and physics-aware learning, robust maps, as well as invertible and constrained neural networks provide unifying challenges for applications across domains, from Medicine to Software Engineering. Addressing them jointly will greatly benefit from the synergies generated in ScaDS.AI Dresden/Leipzig and guide the fundamental algorithmic research with application needs.

Topic Areas

Life Science and Medicine

In the life sciences and medicine, this includes research in applications ranging from personalized therapies over drug discovery and medical image analytics, to providing infrastructures and predictive models for health and biomedical data. Our research provides tailor-made AI-based algorithms and solutions for this. We collaborate with various research institutes in biology, biotechnology, and systems biology, as well as with research hospitals and centers for digital health in order to explore AI technologies within realistic clinical settings.

Earth and Environmental Sciences

Our work in environmental and earth sciences ranges from climate and weather prediction to ecosystem modeling and water resource policy optimization. In all of these application areas, we develop data-driven models, parameter learning, visual computing approaches, and graph learning methods and adapt them to the specific application. We address key data science and AI challenges in these fields, tightly integrated with numerical computation, in collaboration with leading environmental research centers.

Software Engineering

In the application area of software engineering, we consider AI in two ways: First, we research software architectures for implementing scalable and maintainable AI systems. Second, we investigate the use of AI to help design and optimize software systems. This includes AI-based optimization of performance and energy consumption, as well as portability across emerging hardware technologies and novel accelerator designs. For this, we closely collaborate with hardware engineers and software engineers alike, as well as with leading supercomputing installations.

Physics and Chemistry

In physics and chemistry, we research how next-generation data-driven modeling and simulation complement traditional theory-driven approaches. Our research addresses open questions in learning predictive mathematical models for complex dynamics from measurement data, learning molecular structure-function relationships, and providing physically consistent data-driven surrogate models in applications ranging from molecular modeling over astrophysics, quantum physics and photonics to plasma and high-energy physics. We collaborate with a network of physics and chemistry research centers.

Engineering and Business

In the application area of engineering and business, Big Data and AI are paramount for business data analytics, business engineering, virtual prototyping, and traffic/logistics optimization and planning. We focus on intelligent business information systems and AI-based product design. We for example research the intelligent linking of offline analyses with historical data in business intelligence, and we place Big Data in the economical context in collaboration with business engineers and the federal ministry of economics.

Research Focus

Core topics of the applied AI and Big Data research area include data-driven modeling, surrogate models, physics-informed machine learning, and inverse design using AI and Machine Learning approaches suitably adapted to the challenges of the specific application domain. Data Science and Machine Learning methods like sparse multidimensional inference, physics-informed and physics-aware learning, robust maps, as well as invertible and constrained neural networks provide unifying challenges for applications across domains, from Medicine to Software Engineering.