At this year’s ACM Genetic and Evolutionary Computation Conference (GECCO) in Lisbon, Portugal, two papers co-authored by researchers from Mainz University of Applied Sciences were presented, and one of the papers received the Best Paper Award from the Student Workshop. GECCO is considered the largest and most important international conference in the field of metaheuristics and heuristic optimization methods, making this award a significant recognition.
The first paper, titled “Pretraining Reduces Runtime in Denoising Autoencoder Genetic Programming by an Order of Magnitude,” is based on Johannes Reiter’s bachelor’s thesis, which was supervised by Prof. Dr. Dirk Schweim in collaboration with David Wittenberg from the University of Mainz. The research findings were jointly summarized in a paper that was presented at the conference. The aim of the work was to improve the runtime of an innovative algorithm (Denoising Autoencoder Genetic Programming). The algorithm presented in the thesis combines approaches from the fields of deep learning and evolutionary algorithms and achieves a significantly improved runtime through adapted training of the artificial neural network. For the problems examined, the algorithm resulted in the same or even improved solution quality.
In addition, the paper “The Influence of Probabilistic Grammars on Evolution” was presented as a poster. The paper was developed in collaboration with three researchers from the University of Coimbra, Portugal. In terms of content, this paper focuses on the analysis and comparison of various algorithms as well as the identification of opportunities for improvement.
Presenting these papers at GECCO was a valuable opportunity for representatives of Mainz University of Applied Sciences to showcase their research findings to a broad professional audience in an international context and, at the same time, to exchange ideas about the results of other research groups!
