Abstract
Motivated by the desire to better understand the truly remarkable information processing capabilities of the brain, numerous biologically plausible computational models have been explored in the recent decades. Already today, many applications employ neural networks to solve complex real world problems. Significant progress has been made in areas such as speech recognition, robotic controllers, associative memory and function approximation. This book develops an extension for a machine learning technique called the evolving spiking neural network (eSNN). It allows the automatic tuning of the neural and learning-related parameters of eSNN in order to promote its straightforward application to many different problem domains. The book proposes novel evolutionary algorithms capable of efficiently exploring multiple mixed-variable search spaces simultaneously. The enhanced eSNN is comprehensively investigated on benchmark problems and a real-world case study.
