Scientific publications

Highlight papers by the staff of Fraunhofer IKS


Domain Shifts in Reinforcement Learning: Identifying Disturbances in Environments

In their poster and their poster paper Tom Haider and his colleagues present a Markov Decision Process (MDP) to formalize changes in the environment. This allows End-to-End Deep Reinforcement Learning systems to better deal with situations they have not been trained to deal with.


Measuring Ensemble Diversity and Its Effects on Model Robustness

The paper »Measuring Ensemble Diversity and Its Effects on Model Robustness« adresses two questions: To what extent can deep ensembles with the same training conditions differ in their performance and robustness? And are diversity metrics suitable for selecting members to form a more robust ensemble?


Towards Comprehensive Safety Assurance in Cloud-based Systems

In their paper “Towards Comprehensive Safety Assurance in Cloud-based Systems”, Oleg Oleinichenko, Christian Drabek and Anna Kosmalska present a 3-leveled safety analysis process to ensure the dependability of cloud-based systems.


Self-Adaptive Architecture for Federated Learning

In their new paper, Nicola Franco, Hoai My Van, Marc Dreiser and Gereon Weiß propose a self-adaptive architecture for federated learning of industrial automation systems. The approach considers the involved entities on the different levels of abstraction of an industrial ecosystem.


From Black-box to White-box

In their new paper, Franziska Schwaiger and her colleagues analyze the miscalibration of object detection models with respect to image location and box scale. Read more about the results of their experiments in »From Black-box to White-box: Examining Confidence Calibration under different Conditions«.


Cloud-Based Safety-Critical Applications

In their new paper, Christian Drabek and his colleagues identify the main challenges of developing Cyber-Physical Systems-of-Systems (CPSoS) based on several industrial use cases and present their novel approach for designing cloud-based safety-critical applications.


Box Merging Strategies and Uncertainty Estimation Methods

Felippe Schmoeller Roza examines the impact of different box merging strategies for sampling-based uncertainty estimation methods in object detection. The results suggest that estimated variances are a stronger predictor for the detection quality.


Benchmarking Uncertainty Estimation Methods

In this paper »Benchmarking Uncertainty Estimation Methods for Deep Learning With Safety-Related Metrics« which received the Best Paper Award of the SafeAI Workshop the authors compare several state-of-the-art methods for estimating uncertainty for image classifcation with respect to safety-related requirements.


List of publications

Publications by year:



You can access all scientific publications of Frauenhofer IKS and former Fraunhofer ESK in the database Fraunhofer-Publica. You may use the "inst=(iks or esk)" tag in the keyword form to specifically search for publications written by scientists working with our institute.


Gereon Weiß

Contact Press / Media

Dr. Gereon Weiß

Department Head Self-Adaptive Software Systems

Hansastr. 32
80686 Munich

Phone +49 89 547088-348

Karsten Roscher

Contact Press / Media

Karsten Roscher

Departement Head Dependable Perception & Imaging

Hansastr. 32
80686 Munich

Phone +49 89 547088-349

Philipp Schleiß

Contact Press / Media

Philipp Schleiß

Department Head System Safety Engineering

Hansastr. 32
80686 Munich

Phone +49 89 547088-398