DEMYSTIFYING MACHINE LEARNING FOR ARCHITECTURE STUDENTS

Year 2022, - Vol.23 - 16th DDAS (MSTAS) Special Issue -2022, 60 - 67, 23.12.2022

Arzu Sorguç Müge Kruşa Yemişçioğlu Ozan Yetkin

https://doi.org/10.18038/estubtda.1169816

Abstract

With the developments in technology, mass data, new approaches/tools, and the increasing inclusion of machine learning applications, the necessity to teach these concepts and their applications have emerged in all research areas including architecture. In this context, a new course named “machine learning applications in architecture” containing lectures on data, data literacy, patterns, and various kinds of models along with a project conducted by the students was developed and started to be taught in spring’2020. Conducting a class on relatively new subjects for students was a great challenge. Yet, with a well-defined problem-based learning approach, the adaptation of students to the subject took place immediately. It is important to note that as students are equipped with information on machine learning concepts and applications with the given lectures, they were free to choose the project topics of their own which are believed to be one of the reasons for the success of the end results.
As a result of this class, the project topics varied widely as coloring a given painting, predicting the era of a building, interpreting 2d drawings for 3d modeling, optimizing daylight gain, analyzing distinctive features of data in a city, and visualizing data to represent various aspects in data. The outcomes of the class are documented and analyzed to show how information in different fields such as computer science, engineering, statistics, and so on can broaden their thinking of how to attack problems in the architectural design domain. Finally, topics such as data, data literacy, pattern recognition, and intelligent models are projected to play a key role in the future of design education since it provides an interdisciplinary ground to think about problems at hand from a distinct perspective.

Keywords

machine learning, architecture education, data literacy, data-model matching

References

• [1] Khean N, Fabbri A & Haeusler MH. Learning Machine Learning as an Architect, How to? Proceedings of the 36th eCAADe Conference - Volume 1, Lodz University of Technology, Lodz, Poland, 19-21 September 2018; 95-102
• [2] Carpo, M. The Second Digital Turn - Design Beyond Intelligence, MIT Press, Cambridge, 2017
• [3] Mueller C & Danhaive R. Special Subject: Building Technology - Machine Learning for Creative Design | MIT Architecture. Retrieved April 22, 2022, from https://architecture.mit.edu/subject/spring-2019-4s42, 2019
• [4] León, D. A. Artificial Intelligence in Architecture: MaCAD Students Learning How to Radically Innovate the AEC Sector. IAAC. Retrieved April 22, 2022, from https://iaac.net/artificial-intelligence-macad; 2021
• [5] del Castillo y Lopez, JLG. Machine Learning, and the Built Environment. Retrieved April 22, 2022, from https://pll.harvard.edu/course/ai-machine-learning-and-built-environment, 2022
• [6] Miller C. Data Science for Construction, Architecture and Engineering. edX. Retrieved April 22, 2022, from https://www.edx.org/course/Data-Science-for-Construction-Architecture-and-Engineering, 2022
• [7] City of New York. NYC Open Data. Retrieved April 22, 2022, from https://opendata.cityofnewyork.us/data
• [8] Lunchbox. Proving Ground Apps. https://apps.provingground.io/lunchbox. Published June 2, 2021. Accessed May 05, 2022,
• [9] GitHub. https://github.com. Accessed May 05, 2022.