Automated, Explainable Rule Extraction from MAP-Elites Archives

Abstract

Quality-diversity (QD) algorithms that return a large archive of elite solutions to a problem provide insights into how high-performing solutions are distributed throughout a feature-space defined by a user -they are often described as illuminating the feature-space, providing a qualitative illustration of relationships between features and objective quality. However, if there are 1000s of solutions in an archive, extracting a succinct set of rules that capture these relationships in a quantitative manner (i.e. as a set of rules) is challenging. We propose two methods for the automated generation of rules from data contained in an archive; the first uses Genetic Programming and the second, a rule-induction method known as CN2. Rules are generated from large archives of data produced by running MAP-Elites on an urban logistics problem. A quantitative and qualitative evaluation that includes the end-user demonstrate that the rules are capable of fitting the data, but also highlights some mismatches between the model used by the optimiser and that assumed by the user.

Improving the Size and Quality of MAP-Elites Containers via Multiple Emitters and Decoders for Urban Logistics

“…In previous work [27,28,29] we have demonstrated that MAP-Elites can be used provide a diverse set of optimised solutions in a combinatorial optimisation setting, specifically capacitated vehicle routing. While there have been significant developments to the MAP-Elites algorithm since its first introduction, the vast majority have been evaluated in the context of robotics and generative design [15,11,6], using continuous representations.…”

“…While there have been significant developments to the MAP-Elites algorithm since its first introduction, the vast majority have been evaluated in the context of robotics and generative design [15,11,6], using continuous representations. Little attention has been paid to translating these to a combinatorial optimisation setting where there is significant potential for using the family of MAP-Elites algorithms to provide users with increased choice [26,27,29]. We proposed two methods for increasing the solutions returned (s) and diversity (qd) in a combinatorial setting of MAP-Elites: the first used multiple decodings of the same genome, while the second used a range of emitters with a multi-arm bandit to dynamically choose operators to generate new solutions.…”

Applications of Evolutionary Computation

“…In previous work [27,28,29] we have demonstrated that MAP-Elites can be used provide a diverse set of optimised solutions in a combinatorial optimisation setting, specifically capacitated vehicle routing. While there have been significant developments to the MAP-Elites algorithm since its first introduction, the vast majority have been evaluated in the context of robotics and generative design [15,11,6], using continuous representations.…”

“…While there have been significant developments to the MAP-Elites algorithm since its first introduction, the vast majority have been evaluated in the context of robotics and generative design [15,11,6], using continuous representations. Little attention has been paid to translating these to a combinatorial optimisation setting where there is significant potential for using the family of MAP-Elites algorithms to provide users with increased choice [26,27,29]. We proposed two methods for increasing the solutions returned (s) and diversity (qd) in a combinatorial setting of MAP-Elites: the first used multiple decodings of the same genome, while the second used a range of emitters with a multi-arm bandit to dynamically choose operators to generate new solutions.…”