skip to primary navigationskip to content

I-Dec Goh: Objectification of Design Through Machine Learning: An unbiased deep generative framework for architectural design

Supervisor: Dr Ronita Bardhan
profile image i dec goh




Research overview:

Can architectural design be objectified? Detach design from subjectivity such as preference, desire, or aesthetics. Break it down to its core components, such as, what makes a room, occupiable space? A door, an egress? And a window, an aperture? In this society where it is typical for design to supersede architecture for humanity’s greater good, is there a path where we can educate the next generation to create through unbiased design processes? Can we lean such processes towards the scientific reasonings to design instead of the artistic and expressive extractions without compromising the quality of architecture produced? When teaching a child with no prior comprehension of design what architecture is, how would one teach without influencing the child’s ability to create space? Ethically and pragmatically, we cannot experiment with this idea on a human level, but what about a machine? From spoken language to coded language, we can test this idea with machine learning to develop an environment for machine intelligence to create designs. How can we develop a pedagogy for machines to learn design while maintaining objectivity? That is the question to be uncovered in this research.



I-Dec is a designer and a doctoral student at the University of Cambridge. He received his MArch from UC Berkeley and a BSc in Chemistry from NTU Singapore. His research focuses on integrating machine learning with architecture, and his interests in design rely on concepts driven by clarity in form and material.