Research Projects

Recent Research Activities (Selected):


ACADIA 2021 Workshops: Presentations | REALIGNMENTS: Toward Critical Computation

Visit the Workshops Gallery: https://www.artsteps.com/view/6173804dcf1dc439392cc241

The premise of this workshop lies in encoding design intentionality as a continuous set of actions, rather than separate input-output tasks. Workshop participants will be exposed to at least two main artificial neural network structures and the idiosyncrasies of data curation for each type. The goal is to develop a sensibility for such automated processes by leveraging the power of AI tools and also introduce them into ways of evaluating the outcomes from these workflows.
Methods involve experimentation with multiple connected deep learning models, towards prototyping new design workflows. Testing and evaluation of experimental workflows will be pursued within a lens of process-creativity rather than product-creativity.


Latent Morphologies: Disentangling Design Spaces | ACADIA 2021 Workshop

The workshop was offered at the ACADIA 2021 conference. Leaders: Daniel Bolojan, Shermeen Yousif, Emmanouil Vermisso.

The workshop will explore ways to connect different neural networks (i.e. CycleGAN and StyleGAN) to explore the search space of architectural inspiration. Particular semantic references will serve as input for a pre-trained network which outputs data for further investigation using another neural network. The datasets will focus on exploring various resolutions of the urban domain and assess possibilities for emerging patterns, etc through interpolative and extrapolative strategies. From a process point of view we are interested in identifying relevance between certain types of neural networks and their ability to access creative potential in a targeted and/or heuristic/open-ended fashion. Furthermore, it is important to consider the capacity of these nested workflows to alter our immersion in the design investigation by accessing a design space that is otherwise beyond the designer’s reach. By moving beyond rule-based defined design spaces, AI’s feature learning capabilities combined with the incorporation of additional inspirational sources (outside architecture), enable creative exploration within an extended space. Our perception of these expanded design possibilities is crucial because it may point to the direction of future search. 


InclusiveFUTURES Workshop Exhibition Eastern North America

Creative AI Ecologies 2021

In response to recent integration of Artificial Intelligence within architecture, this workshop proposes a rethinking of the architectural design process by introducing nested generative design processes. A new design workflow is offered, for chaining a nested deep learning structure with generative models, to simultaneously address various stages and tasks of the architectural design process. Our approach expands the flexibility of Ai-assisted design, by proposing a series of complementary deep neural networks, establishing a logical continuity in the design decisions while also challenging and augmenting the designer’s agency. This framework encourages the adoption of machine-assisted creativity for tackling various architectural systems, including formal articulation, structural logic, and enclosure responsiveness. Through the combination of parametric and AI models for “representation learning” and “domain-transfer”, parallel iterative workflows address design at the Urban and Architectural scales, using chained supervised and unsupervised neural networks. The instructors are interested to evaluate the impact of multi-designer presence on architectural design, leveraging on the interface among various human design agents.


Deep-Performance

Deep-Performance: Incorporating Deep Learning for Automating Building Performance Simulation in Generative Systems.

Paper presented at the CAADRIA 2021 conference. Authors: Shermeen Yousif and Daniel Bolojan.

In this study, we introduce a newly developed method called Deep-Performance, to enable automatic environmental performance simulation prediction without the need to perform simulations, by integrating deep learning strategies. The aim is to train neural networks on datasets with thousands of building design samples and their corresponding performance simulation. The trained model would offer performance prediction for design options emerging in generative protocols. The research is a work-in-progress within a broader project aimed at automating buildings’ environmental performance evaluations of daylight analysis and energy simulation, using deep learning (DL) models. This paper focuses on the implementation of a supervised DL method for automating the retrieval of daylight analysis metrics, targeting successful daylight design and higher building enclosure efficiency. We have further improved a Pix2Pix model trained on 5 different datasets, each containing 6000 paired images of architectural floor plans and their daylight simulation metrics. In the inference phase, the model was able to accurately predict the daylight simulation for unseen sets of floor plans. For validation, two quantitative assessment metrics were followed to assess the predicted daylight performance against the daylight performance simulation. Both assessment metrics showed high accuracy levels.


Creative AI Ecologies

A research project and a workshop led by: Daniel Bolojan, Shermeen Yousif, Emmanouil Vermisso.

In light of the observed integration of Artificial Intelligence within many industries, this workshop reconsiders the “Architectural Design Cycle”, proposing nested generative AI design processes. Rather than thinking about AI as a “closed” cycle of “input-output”, a series of complementary deep neural networks examine the potential of a logical continuity in AI-driven workflows for architecture, simultaneously challenging and augmenting the designer’s agency. The workshop will deploy AI creativity to tackle a variety of architectural systems, including formal articulation, structural logic, and enclosure responsiveness. Combining parametric and AI layers by means of “domain-transfers” and “representation learning”, two parallel iterative workflows will address design at different scales (Large, Small).


The Role of AI in Architecture Design

A presentation for Research in Action series, Division of Research, Florida Atlantic University.

The session discusses the research and the role of artificial intelligence (AI) in the architecture design process. It covers topics of: the current role of AI, deep learning strategies for an enhanced computational design model, and potentials and challenges of AI creativity for building performance.

%d bloggers like this: