ACQUISITION AND EXPLOITATION OF QUALITATIVE ASPECTS IN 3D SCENE SYNTHESIS
Dimitrios Makris, Georgios Bardis, Georgios Miaoulis, Dimitri Plemenos · International Journal of Artificial Intelligence Tools · 2009
The aim of this paper is to tackle the problem of combinatorial explosion which is inherent in declarative three-dimensional scene synthesis. This is achieved by encoding desired qualitative aspects, not originally supported by the Declarative Modeling methodology, and applying them to the Solution Generation and/or Scene Understanding phase(s), in order to provide the designer with a subset of solutions, which are most representative of the aforementioned aspects. For this reason, we have adapted and applied a machine learning technique, as well as an evolutionary search method, to the current context. The former refers to the gradual construction of a preference model, comprising an incrementally learning mechanism based on actual designer solution evaluation during regular system use. The latter approach models qualitative aspects through a multi-objective genetic algorithm variation based on weighted sums. The algorithm is applied during the generation and understanding phases of Declarative Modeling. The experimental results provide evidence of successful acquisition and exploitation of the desired qualitative characteristics in the paradigm of declarative three-dimensional scene synthesis.