This thesis is aimed at discovering new learning algorithms inspired by principles of biological evolution, which are able to exploit relational and contextual information, viewing clustering and classification problems ...