Dynamical regimes and learning properties of evolved Boolean networks

Boolean networks (BNs) have been mainly considered as genetic regulatory network modelsand are the subject of notable works in complex systems biology literature. Nevertheless, in spite oftheir similarities with neural networks, their potential as learning systems has not yet been fullyinvestigated and exploited.In this work, we show that by employing metaheuristic methods we can train BNs to deal with to twonotable tasks, namely, the problem of controlling the BN's trajectory to match a set of requirementsand the Density Classification Problem. These tasks represent two important categories of problems inmachine learning. The former is an example of the problems in which a dynamical system has to bedesigned such that its dynamics satisfies given requirements. The latter one is a representative task inclassification.We also analyse the performance of the optimisation techniques as a function of the characteristics ofthe networks and the objective function and we show that the learning process could influence and beinfluenced by the BNs' dynamical condition