Due to the progressive growth of data dimensionality, addressing how much data and time is required to train deep learning models has become an important research topic. Thus, in this paper, we present a benchmark for generating floor plans with Conditional Generative Adversarial Networks in which we compare 10 trained models on a dataset of 80.000 samples, the models use different data dimensions and hyper-parameters on the training phase, beyond this objective, we also tested the capability of Convolutional Neural Networks (CNN) to reduce the dataset noise. The models' assessment was made on more than 6 million with the Frétche Inception Distance (FID). The results show that such models can rapidly achieve similar or even better FID results if trained with 800 images of 512x512 pixels, in comparison to high dimensional datasets of 256x256 pixels, however, using CNNs to enhance data consistency reproduced optimal results using around 27.000 images.

Artificial Intelligence represents a substantial part of the available tools on architectural design, especially for Space Layout Planning (SLP). At the same time, the challenge of Mass Customization (MC) is to increase the product variety while maintaining a good cost-benefit ratio. Thus, this research aims to identify new, valid, and easily understandable data patterns through human-machine interaction in an attempt to deal with the challenges of MC during the early phases of SLP. The Design Science Research method was adopted to develop a digital artifact based on deep generative models and a reverse image search engine. The results indicate that the artifact can deliver a series of design alternatives and enhance the navigation process in the solution space, besides giving key insights on dataset design for further research.

This article is part of an exploratory and experimental applied research that seeks to discuss different design strategies with significant potential to stimulate creativity and innovation in the architectural design process. Envisioning a future in which machines are not merely used as tools for creating data, but also to play a role that can enhance the design process itself, this research presents as its fundamental question the possibility of employing a combinatorial use of diffusion models, associated to parametric modeling as a means of predicting, developing, and ultimately optimizing environmentally conscious design proposals. Thus, our ultimate goal is to outline a novel methodology not only capable of stimulating creativity, but also enriching critical thinking and problem-solving skills for sustainable solutions in the early stages of the design process. The strategy here called Design Intelligence Strategy, uses referential design thinking concepts and processes to generate, analyze, compare and (re)systematize data. The object of study is a small house unit with limited constraints, to be implemented in a climatic location through formally adaptive characteristics. The results indicate that the AI generated images have potential to guide the process to climateeffective solutions, besides also being able to be implemented in academic studios.

In the last decades, the initial phase of the design process has been discussed from the point of view of the construction process economy contribution. In this sense, several authors highlight the decisions made at this stage as an impacting point of influence on the direct cost of the building. This concern is repeated in the cases investigated by the ZEMCH (Zero Energy Mass Custom Homes) group in Brazil, which is focused on social housing, consequently, cost becomes a fundamental factor that must be considered on these buildings’ life cycle. Furthermore, ZEMCH’s workshops in Brazil have characterized the initial phase of the design process as a process that requires speed and simultaneous control of a large amount of data by the designers. Over these perspectives, the objective of this work in progress is to verify the potential of a diagrammatic artifact as an intuitive tool of visualizing information and supporting decision making in the initial phase of the urban land subdivision design process, to obtain energy efficiency during the building’s life cycle. Therefore, diagramming information of the environmental conditions into a visualization tool, helps designers to deal with requirements of this phase and cause impact on the operational cost of the building. This research was conducted over the Design Science Research method, and concludes that although visualizing information with the aid of an artifact can play an important role for the practical requirements of the early design phases, the solution can be potentially expanded to other purposes such as teaching, and even for users self-awareness on the operational impact.

Nas últimas décadas os estudos acerca do patrimônio cultural têm interpretado o patrimônio como um fenômeno complexo e multifacetado, no qual ele não é mais entendido como apenas um setor isolado, mas sim como parte integrante da cidade. Nesse contexto, insere-se os ambientes físico-digitais das Smart Cities, que visam aumentar essa conectividade através de dados e interfaces, propondo soluções inovadoras. Nesse sentido, propomos um artefato como interface de interação entre homem-máquina-meio, que identifique as típicas casas de madeira da colonização na região Norte Paranaense, dentro de uma diversidade de estilos arquitetônicos, para que potencialize a percepção do usuário no quesito de continuidade entre a cidade e o patrimônio. Para isso, foi desenvolvido um estudo piloto no qual foi treinado um modelo de Inteligência Artificial para o reconhecimento de padrões dos edifícios com visão computacional. Por fim, os resultados mostram que o modelo do aplicativo obteve uma acurácia compatível com as apresentadas no estado da arte, e apresenta potencial de gerar engajamento dos usuários com o patrimônio, transferência tecnológica e futuro mapeamento de edificações históricas, dados que são potencialmente úteis para a delineação de planos de intervenção na escala macro urbana.