This paper presents an important real-world application of both evolutionary computation and learning, an application to the search for optimal catalytic materials. In this area, evolutionary and especially genetic algorithms are encountered most frequently. However, their application is far from any standard methodology, due to problems with mixed optimization and constraints. The paper describes how these difficulties are dealt with in the evolutionary optimization system GENACAT, recently developed for searching optimal catalysts. It also recalls that the costly evaluation of objective functions in this application area can be tackled through learning suitable regression models of those functions, called surrogate models. Ongoing integration of neural-networks-based surrogate modelling with GENACAT is illustrated on two brief examples. © 2010 Springer-Verlag.
CITATION STYLE
Holeňa, M., Linke, D., & Rodemerck, U. (2010). Evolutionary optimization of catalysts assisted by neural-network learning. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6457 LNCS, pp. 220–229). https://doi.org/10.1007/978-3-642-17298-4_23
Mendeley helps you to discover research relevant for your work.