Modern application scenarios in robot systems, bioinformatics, language processing, etc., but require both efficiency and noise-tolerance and the ability of statistical models and modeling high-level generalization meachanisms structural inference. Therefore various approaches have been proposed to combine the two paradigms. This carefully edited book contains state-of-the-art contributions in neural-symbolic integration, covering `loose 'coupling with the way the kernel structures or recursive models as well as` strong' coupling logic and neural networks.