Integration of convolutional and adversarial networks into building design: A review

Abstract

Convolutional and adversarial networks are found in various fields of knowledge and activities. One such field is building design, a multi-disciplinary and multi-task process involving many different requirements and preferences. Although showing several advantages over traditional computational methods, they are still far from being part of the daily design practice. Nevertheless, if fully integrated, these methods are expected to accelerate design and automate procedures. This paper reviews these methods' latest advances and applications to identify current barriers and suggests future developments. For that, a systematic literature review extended with forward and backward snowball methods was carried out. The focus was on the first design phases, including site layout, floor planning, furniture arrangement, and facade design. The network models show great potential in exploring novel design paths, comparing alternative solutions, and reducing task-associated time and cost. In addition, newer approaches may benefit from combining convolutional and adversarial networks in decision-making since they may complement analysis and synthesis. However, the lack of a smooth integration into the design process and the need for a high-level mastery limit their widespread use. Furthermore, ethical issues arise, such as models being trained with biased datasets, ignoring the intellectual property of the data creators, potential violation of privacy, and the models limiting human creativity.