Distributed embodied evolution for collective tasks
Pedro Trueba, Abraham Prieto, Francisco Bellas · 2013
This paper deals with the formal parametric analysis of an evolutionary paradigm inspired in natural evolution, the distributed Embodied Evolution (EE). The main drawback of this approach is in the high parametric sensitivity it presents, being in addition highly task-dependent. With the aim of understanding the parameter relevance in different fitness landscapes, in this work we present a canonical version of a distributed EE algorithm, propose a set of intrinsic parameters that define its response and analyze them in eight collective problem landscapes.