Networks of evolution
Clodomir Santana, Edward C. Keedwell, Ronaldo Parente de Menezes · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2022
Evolutionary computing has benefited several fields such as psychology, economics, and of course, computer science. These algorithms can tackle challenging real-world optimisation problems using selection, crossover, and mutation operators. Despite the rich literature examining evolutionary algorithms, there are unanswered questions concerning their behaviour and the interplay between operators. This work models genetic algorithms as dynamic networks where nodes represent the population and edges describe the inheritance link between individuals. Using the interaction networks as a proxy, we assessed the impact of different parameters, optimisation problems and operators (i.e. selection, crossover, mutation) on the algorithm's behaviour.