We are very happy that M.Sc. Julian Knaup published his Master Thesis “Impact of Class Assignment on Multinomial Classification Using Multi-Valued Neurons” in a competitive process under guidance of Prof. Dr. Volker Lohweg with Springer Publishers.
The book discusses multilayer neural networks based on multi-valued neurons (MLMVNs) and explicitly addresses the multi-valued neurons (MVNs) in the output layer. Thereby, the classification properties of different class assignment approaches are investigated. Since MVNs require input values located on the unit circle, current transforms were evaluated for their applicability. Subsequently, the developed approaches were evaluated on two datasets.
“Springer awards "BestMasters" to the best master’s theses which have been completed at renowned Universities in Germany, Austria, and Switzerland. The studies received highest marks and were recommended for publication by supervisors. They address current issues from various fields of research in natural sciences, psychology, technology, and economics. The series addresses practitioners as well as scientists and, in particular, offers guidance for early stage researchers." (https://www.springer.com/series/13198)