Real Application Challenges in Evolutionary Optimization? People!
Allgemeines
Art der Publikation: Conference Paper
Veröffentlicht auf / in: Applications of Evolutionary Computation. EvoApplications 2025
Jahr: 2025
Seiten: 469-481
Veröffentlichungsort: Cham
Verlag (Publisher): Springer
DOI: https://doi.org/10.1007/978-3-031-90062-4_29
Autoren
Tobias Rodemann
Zusammenfassung
The application of evolutionary optimization methods for real world problems is often far less straight-forward than expected. One of the main challenges are the people involved in real business situations that decide on whether optimization projects are successful or not. In this work we present a few insights from years of applying Evolutionary Algorithms (mostly multi- and many-objective) with in-house customers and point to some key psychological insights that can explain several of the major issues that we encountered in our work but are also reported repeatedly in application sessions on major conferences. We argue that more convincing sales messages are needed; and explain, why trust, not better objective values, might be the ultimate goal of any optimization process. We can’t provide numbers or new algorithms, but maybe a different perspective on some of the most persistent practical challenges.