Designing and evaluating algorithms for automated discovery of adaptive network models based on generative network automata

Generative Network Automata (GNA) is a powerful tool for the study of adaptive networks. It has the ability to represent a wide range of dynamics by leveraging its inherent generality. The ability to automatically discover underlying dynamics of adaptive network input has been theoretically proposed using GNA. This work tries to answer the question as to whether it is possible to create a practical implementation of GNA for the automatic discovery of dynamical rules that capture the state transition and topological transformation of complex adaptive networks. The results show that our algorithms and software (called PyGNA) correctly identifies the dynamics of a set of simple adaptive networks. Capturing the dynamics of more complex adaptive networks remains a challenge that will require further algorithm improvement.