Browsing by Author "Yildiz, Berfin"

Now showing 1 - 3 of 3

Citation - WoS: 2
Citation - Scopus: 2
Architectural space classification considering topological and 3D visual spatial relations using machine learning techniques
(Routledge, 2024) Berfin Yıldız; Gulen Cagdas; Ibrahim Zincir; Yildiz, Berfin; Cagdas, Guelen; Zincir, Ibrahim
The paper presents a novel method for classifying architectural spaces in terms of topological and visual relationships required by the functions of the spaces (where spaces such as bedrooms and bathrooms have less visual and physical relationships due to the privacy while common spaces such as living rooms have higher visual relationship and physical accessibility) through machine learning (ML). The proposed model was applied to single and two-storey residential plans from the leading architects of the 20th century Among the five different ML models whose performances were evaluated comparatively the best results were obtained with Cascade Forward Neural Networks (CFNN) and the average model success was calculated as 93%. The features affecting the classification models were examined based on SHAP values and revealed that width control 3D visibility and 3D natural daylight luminance were among the most influential. The results of five different ML models indicated that the use of topological and 3D visual relationship features in the automated classification of architectural space function can report very high levels of classification accuracy. The findings show that the classification model can be an important part of developing more efficient and adaptive floor plan design building management and effective reuse strategies. © 2024 Elsevier B.V. All rights reserved.
Automated Two-Story Housing Floor Plan Generation Using Generative Adversarial Networks
(Springer International Publishing AG, 2025) Yildiz, Berfin; Cagda, Gillen; Zincir, Ibrahim
Automating the generation of two-story housing floor plans has emerged as a significant area of focus in architectural design research, driven by the need to enhance efficiency, creativity, and functionality in the design process. This study introduces a GAN-based framework for the automated generation of two-story housing layouts, incorporating architectural constraints such as functional zoning, multi-level connectivity, open-plan configurations, and visual relationships. By leveraging advanced deep learning techniques, the proposed framework achieves a balance between design creativity and practical functionality, addressing the unique challenges posed by multi-level spatial arrangements. The results demonstrate the model's ability to generate diverse and coherent floor plans that effectively meet the complexities of two-story layouts. This research underscores the transformative potential of deep learning models in architectural design, while acknowledging existing limitations in managing multi-level spatial relationships and user interaction. With continued advancements, AI has the potential to play a pivotal role in supporting architects-optimizing workflows, enabling creative exploration, and fostering user-centered, innovative designs. Ultimately, this work sets the stage for further progress in automated multi-story housing design, paving the way for a more collaborative and technology-driven architectural future.
Citation - WoS: 22
Citation - Scopus: 30
Fuzzy logic in agent-based modeling of user movement in urban space: Definition and application to a case study of a square
(Elsevier Ltd, 2020) Berfin Yıldız; Gulen Cagdas; Yildiz, Berfin; Cagdas, Gulen
The growing complexity of design processes increases the distance between designer and user which makes it challenging to consider user experience in design. Computational models can help us to simulate user behaviors where agents represent users as a collection of autonomous decision-making entities. In this context development of these models supports early stage decision-making in urban design. The aim of this study is to investigate how the user is involved in urban space and to analyze the relationship between urban space components and the users’ movement to be able to develop a model for user movement simulation. This paper follows a five-step consecutive process: (1) data collection with observation studies and environmental analysis (2) interpretation of the data using fuzzy logic (3) agent-based model development (4) model implementation (5) evaluation and validation. The interpretation of the observation data is to calculate the attractiveness value of urban space components with fuzzy logic. The value is then defined as attract force on agent-based simulation model. The simulation results are evaluated comparatively using observation outputs. As a case study for the model capabilities demonstration a square is chosen (Konak Square Izmir Turkey). Two models for morning and evening timelines are defined and tested to be able to simulate user movement in the square. Thereafter the efficiency of the model is examined by comparing the simulation results and observation data by the Mean Absolute Percentage Error (MAPE) and Secant Cosine Calculation methods. © 2019 Elsevier B.V. All rights reserved.