Best Paper Award at ESWEEK 2023’s MEMOCODE 2023 in Hamburg, 22.09.2023
The paper “Hybrid Genetic Reinforcement Learning for Generating Run-Time Requirement Enforcers” (Jan Spieck, Pierre-Louis Sixdenier, Khalil Esper, Stefan Wildermann and Jürgen Teich) has won the Best Paper Award at ESWEEK 2023’s 21st ACM-IEEE International Symposium on Formal Methods and Models for System Design (MEMOCODE) in Hamburg, Germany.
In their work, the authors propose a novel design space exploration methodology that uses a combination of evolutionary algorithms and reinforcement learning to efficiently find verifiable feedback-based controllers for non-functional requirements such as latency or energy. The approach is shown to significantly reduce exploration time and to outperform existing methods in generating controllers with higher probabilities of meeting verification goals in three case studies. The paper originated as part of TRR 89 Invasive Computing.