With a rich 90-year history of innovation, the Fraunhofer Heinrich Hertz Institute (HHI) is one of the oldest artificial intelligence research laboratories in the world. Based in Berlin, Germany, the HHI researches the impact of technological advances on our future, especially in the way we communicate.

HHI's machine learning group focuses on deep learning (Deep Learning) research. Recently, the AI research institute was testing 5G technology in Berlin.

Research Areas

The main focus of the Artificial Intelligence department are the theoretical and algorithmic foundations of machine learning, the development of novel methods and models of deep learning and the application of these AI techniques in practice. A core topic is the explainability and interpretability of deep learning models. With LRP and SpRAy, the department has developed pioneering algorithms for the explanation of deep neural networks, which have already been used successfully in various scientific and industrial applications. The development of methods for compression of neural networks and for efficient federated learning are two further central research topics of the department. 


Testing and Experimentation Facility for Health AI and Robotics

the research team is developing a test infrastructure (both virtual and physical) that can evaluate various technologies in realistic environments, including hospitals and laboratories. For instance, software for patient care or diagnostics, as well as surgical or nursing robots can be tested by users.

Reconfigurable Superconducting and Photonic Technologies of the Future

Computing with light using integrated optics has seen huge progress over the last 3-4 years in multiple fields such as neuromorphic computing, quantum computing, and on-chip data storage. This has created a vast ecosystem that relies on high-speed reconfigurations of nano-photonic circuits (such as their use as synapses or in routing applications) and ultra fast yet high-resolution, low-power photo detection.

Synthesizing realistic variations in data for reliable medical machine learning at scale

Due to the increasing burden on the healthcare system, artificial intelligence (AI)-based algorithms are currently being developed in order to make medical workflows more efficient.To generate the medical image data, this project will use and further develop deep-learning based generative AI algorithms (GANs), which learn and realistically replicate image features. 

Quantifying Uncertainties in Explainable AI

Thus far, most deep learning studies have been empirically driven and are usually viewed as "black boxes": they can produce a decision but the grounds for this decision are unclear. In MATH+, a theoretical understanding of the explainability of deep neural networks is developed. For a given decision, the features of input data that played the largest role are identified and the associated uncertainties of a decision are quantified.

Transparent Medical Expert Companion

The quantity of data in the health sector (e.g., imagery and ECG time series) is growing exponentially. Machine learning can support the analysis and interpretation of these data so that medical practitioners can create diagnoses more efficiently. In TraMeExCo, the robustness and transparency of diagnostic prediction through machine learning is explored. For two clinical fields (pathology and pain analysis), machine learning methods are tested with three types of data (microscopy images, pain videos, and ECG time series).

Data- and AI-supported early warning system

The aim of DAKI-FWS is to produce an early warning system based on various data sources and leveraging novel methods (including artificial intelligence), which will strengthen economic resilience in Germany. The early warning system (EWS) will integrate key data sources but remain modular, allowing different sectors to use the system to mitigate the impacts of hazards.