Machine Learning, Artificial Intelligence, and Urban Assemblages

Recent advances in generative design, simulation, computational optimisation, and machine learning opened up possibilities in employing data-driven workflows to achieve design solutions of unprecedented performance. This chapter showcases generative methods for urban spatial configurations that integrate a number of different simulation engines, along with InFraRed, into one framework. This allows us to quickly explore thousands of urban design alternatives by generating a diverse and informative design and performance data set. Conceptional InFraRed is rooted in the paradigm of cognitive design computing, which emphasises an almost symbiotic relationship between computational methods and human interaction. Performance evaluation methods of urban spaces, either by generating synthetic data or using real-world data, becomes more meaningful and interpretable when correlated with design configuration indicators. The InFraRed web platform currently offers a few selected services; however, further services will be included in the near future.