The AAAI Conference on Artificial Intelligence is one of the top conferences on Artificial Intelligence research. Our paper Interpreting Neural Networks as Quantitative Argumentation Frameworks explains how subsymbolic neural networks can be understood as symbolic argumentation frameworks. This connection allows using learning algorithms for neural networks to learn argumentation frameworks from data and incorporating human-readable background knowledge in form of argumentation frameworks into deep learning architectures. More fundamentally, our paper shows how multilayer perceptrons can be generalized to cyclic network structures and that the resulting values are meaningful from a common-sense (argumentation) perspective.