Ata Kaban

Ata Kaban is Lecturer and part of the Natural Computation Group at the School of Computer Science, University of Birmingham.

Generative design and aestetic perception
Objectives include understanding / exploring of the following:-

• what are the deterministic vs. random components in artistic creativity and aestetic perception
• how do we represent / perceive the style of an aestetic object. I am interested in investigating a modelling hypothesis related to minimum complexity representations: To what extent specific styles are represented compactly as specific combinations (out of a continuum of possibilities) of a (possibly infinite) set of common stylistic primitives and how can this be captured in an identifyable model. This would also allow us to control stylistic similarity / dissimilarity relations as well as novelty of stylistic components in a more objective manner.
• what are the structural differences & similarities between visual and auditive stylistic perception. Building on the previous item, we may comparatively study the characteristics of both the proportions and the forms of combination which stylistic primitives typically exhibit in the case of these two best established different forms of artistic communication.

This is intended to be complementary to both discovery based generation of aestetic objects (such as evolutionary techniques) and personalisation of the design for individual users. The outlined stylistic analysis approach tries to capture the diversity structure of stylistic expression that is a dimension orthogonal to that of user identity. (an illustrative analogy in this respect are the complementary tasks of face expression recognition vs. recognition of person identity from a set of face images) It can be applied either globally for all users or individually for each user or collaboratively. Joint models of stylistic structure and user satisfaction function can also be developed.

---