Intelligently combining design features of different designs can yield radically new functional designs.

Using Deep Learning to Generate Conceptual Design


Collaborators: Imdat As + Prithwish Basu

Project Duration: Jan 2018 - May 2018

Funding: DARPA (Defense Advanced Research Projects Agency) Grant

Researchers: 1

Status: Complete



Artificial intelligence, and in particular machine learning, is a fast-emerging field. Research on artificial intelligence focuses mainly on image-, text- and voice-based applications, leading to breakthrough developments in self-driving cars, voice recognition algorithms and recommendation systems. In this project we worked on an alternative graph-based machine learning system that deals with three-dimensional space, which is more structured and combinatorial than images, text or voice. Specifically, we explored a function-driven deep learning approach to generate conceptual design. We trained and used deep neural networks to evaluate existing designs encoded as graphs, extract significant building blocks as subgraphs and merge them into new compositions. Finally, we explored the application of generative adversarial networks to generate entirely new and unique designs.




  • 107th ACSA Conference, “Composing Frankensteins: Data-driven design assemblies through graph-based deep neural networks,” Computer Composition: Design after Machine Learning, Carnegie Mellon University, Pittsburgh, PA, March 28-30, 2019.

  • Hochschule Wismar, University of Applied Sciences, Technology, Business and Design, “24 hours Berlin City West,” International Workshop on the Future City in Berlin, Germany, March 16-24, 2019.

  • Bilkent University, “The Future is History: Architecture in the Age of Artificial Intelligence,” Ankara, Turkey, 2018.

Topological summary of latent design rules of architectural data generating automated designs.

Discovering latent building blocks in design data using deep learning and topological data analysis.