Machine Learning in Architecture

Abstract: The development of image-based machine learning technology brings new ways of how we use and produce images. Due to the significant reliance on images and drawings, machine learning technology might achieve new potentials for the architecture discipline. My thesis explored these potentials through conducting experiments. In the first part of my thesis, I used the image-based machine learning algorithm to conduct facade and church generation experiments to discover the relationship between architect, image, and algorithm. In the second part, I evaluated the quality of generated results and considered the future potential of using machine learning technology in architecture.