Karl Meinke

Karl Meinke is a Professor at KTH Royal Institute of Technology, where he serves as Head of the Computer Science Department and Head of the Division of Theoretical Computer Science within the School of Electrical Engineering and Computer Science. His research focuses on applying machine learning techniques to software testing, particularly for safety-critical systems like autonomous vehicles and embedded systems.

His research interests span machine learning, software testing, safety critical systems, embedded systems, autonomous driving, digital pathology, and graph neural networks. Meinke has developed innovative approaches like Learning-Based Testing that combine machine learning with formal methods for system validation. His work bridges theoretical computer science with practical applications in automotive systems and medical diagnostics.

His recent publications show a strong trend toward applying graph neural networks to diverse domains including program analysis, digital pathology, and autonomous vehicle testing. His research demonstrates a consistent focus on solving the test oracle problem and generating meaningful test cases for complex systems where traditional testing approaches fall short.

Meinke actively collaborates with Karolinska Institutet (KI), indicating interdisciplinary work between computer science and medical research. He is responsible for Masters level education in software testing at KTH and serves as examiner for several advanced courses including Degree Projects in Computer Science and Software Reliability.

His research group has developed tools like LBTest for learning-based testing of reactive systems, and he has secured funding for projects such as the ITEA3 Testomat Project focused on next-level test automation. His work has significant implications for validating autonomous systems where safety is paramount.

Meinke leads research in using machine learning to address fundamental challenges in software testing, particularly for systems where traditional test oracles are unavailable or impractical. His approach of combining active learning with formal specifications has created new pathways for validating complex cyber-physical systems.