News from the Fraunhofer Institute for Cognitive Systems IKS

The 41st international SafeComp conference (Safety, Reliability and Security of Computer-based Systems), which drew to an end in Munich on September 9, gave 156 experts from 18 countries an opportunity to focus intensively on is-sues regarding the safety and reliability of computer-based systems. Next year, SafeComp will take place in Toulouse.

On June 28, a Canadian delegation of parliamentarians visited Fraunhofer IKS together with two members of the Bavarian Parliament. In addition to a presentation on Fraunhofer IKS by Director Mario Trapp, the delegation was given a tour in the Living Lab and got insights into various research fields and projects of Fraunhofer IKS, including AI assurance, quantum computing and adaptive resilience management.

Dr. Narges Ahmidi, Head of Reasoned AI Decisions, explains in this video, how artificial intelligence and machine learning will share medical technology in the near future. AI is particularly helpful where a lot of data is involved. The areas of application range from diagnosis, for example through image analysis, to the treatment of patients and the maintenance of medical equipment in intensive care units.

Fraunhofer IKS is now part of the Quantum Business Network (QBN). The QBN brings together people from science, industry and politics to shape the future of quantum technologies together. The goal is technological progress through industry collaborations and knowledge transfer.

In the future, quantum computers will be able to solve highly complex tasks that traditional computers have failed to do so far. In the QuaST project, which is funded by Germany’s Federal Ministry of Economic Affairs and Climate Action, seven partners from science and industry are researching how to make quantum computing as easily accessible as possible for companies.

With the flexible ezCar2x® software framework Fraunhofer IKS provides the key components needed to rapidly create prototype applications for networking vehicles. The link to the complete source code and the corresponding manual can be found on the ezCar2x® website.

Fraunhofer IKS brings together safety and artificial intelligence (AI): A new application assesses the reliability of artificial intelligence at the push of a button. The application tests the robustness of the results of their AI model. The new platform uses metrics for uncertainty and robustness quantification to assess the reliability of the analyses of the tested AI models.

This online seminar looks at the role that Artificial Intelligence (AI) and Machine Learning (ML) currently play in safety-critical systems in rail transport and what we can expect in the future - for example, in autonomous traction vehicles (automatic train operation) or in the field of autonomous maintenance of rail vehicles and rail infrastructure.

The 41st International Conference on Computer Safety, Reliability and Security, SafeComp, will take place next year in Munich. For its 2022 edition, the annual event has the motto »New frontiers of safety assurance«. It will be hosted by the Fraunhofer Institute for Cognitive Systems IKS and Friedrich-Alexander-Universität Erlangen-Nuremberg (FAU).  

After the successful premiere in 2021, the Fraunhofer Institute for Cognitive Systems will again host the international conference »Safetronic« 2022. The event is planned for November 8 and 9, 2022.

It is essential for safety-critical applications of deep neural networks to determine when new inputs are significantly different from the training distribution. This paper explores this out-of-distribution (OOD) detection problem for image classification using clusters of semantically similar embeddings of the training data and exploit the differences in distance relationships to these clusters between in- and out-of-distribution data.

Model compression is widely employed to deploy convolutional neural networks on devices with limited computational resources or power limitations. For high stakes applications, it is, however, important that compression techniques do not impair the safety of the system. This paper investigates the changes introduced by three compression methods that go beyond the test accuracy.

A serious problem in image classification is that a trained model might perform well for input data that originates from the same distribution as the data available for model training but performs much worse for out-of-distribution (OOD) samples. This paper proposes a first-of-its-kind OOD detection architecture named OODformer that leverages the contextualization capabilities of the transformer.

Cloud technologies is one of the more recent additions to the common portfolio of software architects, which in general decouple software and the resources required for running. This allows the software services to be quickly scaled according to demand. Nevertheless, there are several pitfalls that can be encountered when integrating cloud technologies into your software architecture.

Striving for excellence in scientific research on trustworthy Artificial Intelligence (AI), Dr. Narges Ahmidi also aims to encourage young female scientists to enter science and become leaders.

There are still a number of technical and regulatory obstacles to overcome before autonomous cars are ready for mass production. But when it comes to rail transport, things are much further ahead. Self-driving trains are already a common mode of transport. But how does autonomous driving on rails work?