Artificial Creativity of Evolutionary Swarm Systems

We present an artificial creative system that autonomously keeps producing various spatial patterns using evolutionary swarm systems and an automated object harvesting mechanism. This system consists of two concurrent, asynchronous computational processes: (1) A long-term simulation process of the evolutionary Swarm Chemistry model made of 10,000 self-propelled particles moving in a finite, periodically perturbed two-dimensional continuous space in which the behavioral rules of particles are transmitted (and possibly mutated) among them through collision, and (2) an image analysis process that analyzes the snapshots of simulations generated by the first process and automatically extracts and saves objects (spatially contiguous clusters of particles) from those snapshots. In this system, there is no globally defined fitness function or selection criterion for “organisms” to adapt to. Every evolutionary and self-organizing dynamics takes place in a fully bottom-up manner, which makes this system highly unique compared to other more conventional evolutionary algorithms. As a result, this system can and does produce a wide variety of unprescribed, often quite creative, spatial patterns, some of which may appear very like-like. We discuss the creative power of such distributed artificial life systems and their potential implications for the open-endedness of evolutionary processes in general.

Hiroki Sayama
Tuesday, September 25, 2018 - 12:00 to 12:15


The official Hotel of the Conference is
Makedonia Palace.

Conference Organiser: NBEvents

The official travel agency of the Conference is: Air Maritime

Photo of Thessaloniki seafront courtesy of Juli Bellou
fb flickr flickr