Computational Intelligence Approaches to Computational Aesthetics

Computational aesthetics is an area of research which attempts to develop computational methods that can perform aesthetic judgements in the same way as humans (Hoenig, 2005). It is an area of research which has not developed as a separate discipline till relatively recently. The notion of aesthetics is highly intuitive and often subjective. An aesthetic experience can be negative, positive or more subtly nuanced. Human beings have a strong and deep sense of aesthetics, however rationalising aesthetic decisions is challenging. As such developing computational models to make aesthetic decisions is particularly challenging. While computational intelligence techniques such as evolutionary algorithms have been able to solve many real world challenges, still such techniques are not widely used to solve problems that involve aesthetic decisions. Making an aesthetic decision often requires a human in the loop which in turn creates a barrier between computational intelligence and aesthetics. However recent advancements in computational aesthetics have made computer generated art and aesthetics realisable in several domains (den Heijer & Eiben, 2012; DiPaola & Gabora, 2009). The purpose of this article is to summarise the advancements in the area of computational aesthetics, challenges involved, computational intelligence approaches to art and aesthetics and possible future directions. The article first summarises early attempts to define aesthetics, through to more contemporary definitions and attempts in developing computational models of aesthetics in various domains. Then, it highlights the challenges associated with bridging the gap between aesthetics and computational intelligence. Thereafter it discusses how computational intelligence techniques are used in art and aesthetics ranging from simple classification problems to more advanced problems such as automatic generation of art artefacts, stories and simulations. The article concludes highlighting the future research directions that need to be undertaken in order to make significant advancements in computational aesthetics and its applications.