Skavara, M.M.E.; (2009) Learning emergence: adaptive cellular automata façade trained by artificial neural networks. Masters thesis , UCL (University College London).

Preview

PDF 19042.pdf Download (4MB)

Abstract

This thesis looks into the possibilities of controlling the emergent behaviour of Cellular Automata (CA) to achieve specific architectural goals. More explicitly, the objective is to develop a performing, adaptive building facade, which is fed with the history of its achievements and errors, to provide optimum light conditions in buildings’ interiors. To achieve that, an artificial Neural Network (NN) is implemented. However, can an artificial NN cope with the complexity of such an emergent system? Moreover, can such a system be trained to compute and yield patterns with specific regional optima, using simple inputs deriving from its environment? Both Backpropagation and optimisation using Genetic Algorithms (GA) are tested to reassign the weights of the network and several experiments are conducted regarding the structure and complexity of both CA and NN. Here it is argued that in fact, it is possible to train such a system although the level of success is strongly dependent on the degree of complexity and the level of resolution and accuracy. By taking advantage of the structural attributes of certain CA that go beyond just a higher order stability, this dissertation suggests that such an evolutionary, computational approach can lead to adaptive and performative architectural spaces of high aesthetic value.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item