Conflict Resolution and a Framework for Collaborative Interactive Evolution

Proceedings of the AAAI Conference on Artificial Intelligence, 21

Volume

Technical Papers

Machine Learning

Interactive evolutionary computation (IEC) has proven useful in a variety of applications by combining the subjective evaluation of a user with the massive parallel search power of the genetic algorithm (GA). Here, we articulate a framework for an extension of IEC into collaborative interactive evolution, in which multiple users guide the evolutionary process. In doing so, we introduce the ability for users to combine their efforts for the purpose of evolving effective solutions to problems. This necessarily gives rise to the possibility of conflict between users. We draw on the salient features of the GA to resolve these conflicts and lay the foundation for this new paradigm to be used as a tool for conflict resolution in complex group-wise human-computer interaction tasks.