AI in architectural design: Adaptive tabular model for classifying intelligent tools
(2024) International conference AI & Automation in Architectural Research and Practice (AIARP 2024) — Location: ENSA Paris Malaquais, Paris (10.October.2024)
Files
102_241010_AIARP2024-Adaptive-tabular-model.pdf
Open Access - Adobe PDF
- 1.02 MB
Details
- Authors
- Abstract
- This research proposes an adaptive model in the form of a periodic table to classify intelligent tools used in architectural design. Inspired by Mendeleev's periodic table, this model aims to organize various artificial learning algorithms based on their complexity, learning method, and the phases of artificial intelligence (AI) processes. The objective is to make the functions and potential of design assistance tools accessible to non-experts while facilitating the understanding of possible combinations between these algorithms. The proposed periodic table is structured around eight periods representing increasing levels of algorithmic complexity, six families based on learning methods and their depth, and four blocks representing the typical phases of data processing: analysis, model development, decision-making, and causal inference. Each algorithm is identified by a label that provides information about the types of inputs and outputs it processes, its accuracy, its learning duration, and whether it models linear or nonlinear relationships. This categorization allows for a better understanding and comparison of various algorithms in relation to their specific applications in architectural design, where AI does not replace human cognition but enriches and extends it.
- Affiliations
Citations
Roobaert, L., & Claeys, D. (2024). AI in architectural design: Adaptive tabular model for classifying intelligent tools. International conference AI & Automation in Architectural Research and Practice (AIARP 2024), ENSA Paris Malaquais, Paris.