Cumincad : CUMINCAD Papers : Search Results

CumInCAD is a Cumulative Index about publications in Computer Aided Architectural Design
supported by the sibling associations ACADIA, CAADRIA, eCAADe, SIGraDi, ASCAAD and CAAD futures

PDF papers
References

Hits 1 to 20 of 525

Reformat results as: short short into frame detailed detailed into frame

_id	ecaadesigradi2019_171
id	ecaadesigradi2019_171
authors	Uzun, Can and Çolako?lu, Meryem Birgül
year	2019
title	Architectural Drawing Recognition - A case study for training the learning algorithm with architectural plan and section drawing images
doi	https://doi.org/10.52842/conf.ecaade.2019.2.029
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 2, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 29-34
summary	This paper aims to develop a case study for training an algorithm to recognize architectural drawings. In order to succeed that, the algorithm is trained with labeled pixel-based, architectural drawing (plan and section) dataset. During the training process, transfer learning (pre-training model) is applied. The supervised learning and convolutional neural network are utilized. After certain iterations, the algorithm builds awareness and can classify pixel-based plan and section drawings. When the algorithm is shown a section that is not produced with conventional drawing technic but through hybrid technics, it could predict the drawing class correctly with %80 of accuracy. On the other hand, some of the algorithm prediction is misoriented. We examined this prediction problem in the discussion section. The results illustrate that neural networks are successful in training algorithms to recognize and classify pixel-based architectural drawings. But for a highly accurate algorithm prediction, the dataset of the drawing images must be ordered, according to sample resolution, sample size and sample coherence for the dataset.
keywords	Classification Algorithm; Pixel-Based Architectural Drawing Recognition; Plan; Section
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:57

_id	ecaadesigradi2019_514
id	ecaadesigradi2019_514
authors	de Miguel, Jaime, Villafa?e, Maria Eugenia, Piškorec, Luka and Sancho-Caparrini, Fernando
year	2019
title	Deep Form Finding - Using Variational Autoencoders for deep form finding of structural typologies
doi	https://doi.org/10.52842/conf.ecaade.2019.1.071
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 1, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 71-80
summary	In this paper, we are aiming to present a methodology for generation, manipulation and form finding of structural typologies using variational autoencoders, a machine learning model based on neural networks. We are giving a detailed description of the neural network architecture used as well as the data representation based on the concept of a 3D-canvas with voxelized wireframes. In this 3D-canvas, the input geometry of the building typologies is represented through their connectivity map and subsequently augmented to increase the size of the training set. Our variational autoencoder model then learns a continuous latent distribution of the input data from which we can sample to generate new geometry instances, essentially hybrids of the initial input geometries. Finally, we present the results of these computational experiments and lay out the conclusions as well as outlook for future research in this field.
keywords	artificial intelligence; deep neural networks; variational autoencoders; generative design; form finding; structural design
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:55

_id	caadria2019_245
id	caadria2019_245
authors	Jiaxin, Zhang, Yunqin, Li, Haiqing, Li and Xueqiang, Wang
year	2019
title	Sensitivity Analysis of Thermal Performance of Granary Building based on Machine Learning
doi	https://doi.org/10.52842/conf.caadria.2019.1.665
source	M. Haeusler, M. A. Schnabel, T. Fukuda (eds.), Intelligent & Informed - Proceedings of the 24th CAADRIA Conference - Volume 1, Victoria University of Wellington, Wellington, New Zealand, 15-18 April 2019, pp. 665-674
summary	The granary building form has significant effects on thermal performance, especially in hot climate regions. This research is focused on exploring the influences of parameters relevant to building form design on thermal performance for granary buildings in Jiangsu and Anhui, China(both provinces belong to the hot summer region). The usual method is to use simulation software to perform a sensitivity analysis of thermal performance to assess the impacts of granary design parameters and identify the essential characteristics. However, many factors are affecting the thermal performance of granary buildings. The use of traditional energy simulation software requires calculation and analysis of a large number of models. In this study, we build a machine learning model to predict the thermal performance of granary buildings and identify the most influential design parameters of thermal performance in granary building. The input parameters include outdoor temperature, building height, aspect ratio, orientation, heat transmission coefficient of the wall and roof, and overall scale. The results show that the overall building scale is the most influential variable to the annual electricity consumption for cooling, whereas the heat transmission coefficient of the roof is the most influential to the change of the indoor temperature.
keywords	Sensitivity analysis; Artificial Neural Networks (ANNs); Thermal performance; Granary building
series	CAADRIA
email
last changed	2022/06/07 07:52

_id	ecaadesigradi2019_339
id	ecaadesigradi2019_339
authors	Kinugawa, Hina and Takizawa, Atsushi
year	2019
title	Deep Learning Model for Predicting Preference of Space by Estimating the Depth Information of Space using Omnidirectional Images
doi	https://doi.org/10.52842/conf.ecaade.2019.2.061
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 2, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 61-68
summary	In this study, we developed a method for generating omnidirectional depth images from corresponding omnidirectional RGB images of streetscapes by learning each pair of omnidirectional RGB and depth images created by computer graphics using pix2pix. Then, the models trained with different series of images shot under different site and weather conditions were applied to Google street view images to generate depth images. The validity of the generated depth images was then evaluated visually. In addition, we conducted experiments to evaluate Google street view images using multiple participants. We constructed a model that estimates the evaluation value of these images with and without the depth images using the learning-to-rank method with deep convolutional neural network. The results demonstrate the extent to which the generalization performance of the streetscape evaluation model changes depending on the presence or absence of depth images.
keywords	Omnidirectional image; depth image; Unity; Google street view; pix2pix; RankNet
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:52

_id	ecaadesigradi2019_135
id	ecaadesigradi2019_135
authors	Newton, David
year	2019
title	Deep Generative Learning for the Generation and Analysis of Architectural Plans with Small Datasets
doi	https://doi.org/10.52842/conf.ecaade.2019.2.021
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 2, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 21-28
summary	The field of generative architectural design has explored a wide range of approaches in the automation of design production, but these approaches have demonstrated limited artificial intelligence. Generative Adversarial Networks (GANs) are a leading deep generative model that use deep neural networks (DNNs) to learn from a set of training examples in order to create new design instances with a degree of flexibility and fidelity that outperform competing generative approaches. Their application to generative tasks in architecture, however, has been limited. This research contributes new knowledge on the use of GANs for architectural plan generation and analysis in relation to the work of specific architects. Specifically, GANs are trained to synthesize architectural plans from the work of the architect Le Corbusier and are used to provide analytic insight. Experiments demonstrate the efficacy of different augmentation techniques that architects can use when working with small datasets.
keywords	generative design; deep learning; artificial intelligence; generative adversarial networks
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:58

_id	caadria2019_126
id	caadria2019_126
authors	Ng, Jennifer Mei Yee, Khean, Nariddh, Madden, David, Fabbri, Alessandra, Gardner, Nicole, Haeusler, M. Hank and Zavoleas, Yannis
year	2019
title	Optimising Image Classification - Implementation of Convolutional Neural Network Algorithms to Distinguish Between Plans and Sections within the Architectural, Engineering and Construction (AEC) Industry
doi	https://doi.org/10.52842/conf.caadria.2019.2.795
source	M. Haeusler, M. A. Schnabel, T. Fukuda (eds.), Intelligent & Informed - Proceedings of the 24th CAADRIA Conference - Volume 2, Victoria University of Wellington, Wellington, New Zealand, 15-18 April 2019, pp. 795-804
summary	Modern communication between built environment professionals are governed by the effective exchange of digital models, blueprints and technical drawings. However, the increasing quantity of such digital files, in conjunction with inconsistent filing systems, increases the potential for human-error upon their look-up and retrieval. Further, current methods are manual, thus slow and resource intensive. Evidently, the architectural, engineering and construction (AEC) industry lacks an automated classification system capable of systematically identifying and categorising different drawings. To intercede, we aim to investigate artificially intelligent solutions capable of automatically identifying and retrieving a wide set of AEC files from a company's resource library. We present a convolutional neural network (CNN) model capable of processing large sets of technical drawings - such as sections, plans and elevations - and recognise their individual patterns and features, ultimately minimising laboriousness.
keywords	Convolutional Neural Network; Artificial Intelligence; Machine Learning; Classification; Filing architectural drawings.
series	CAADRIA
email
last changed	2022/06/07 07:58

_id	ecaadesigradi2019_602
id	ecaadesigradi2019_602
authors	Toulkeridou, Varvara
year	2019
title	Steps towards AI augmented parametric modeling systems for supporting design exploration
doi	https://doi.org/10.52842/conf.ecaade.2019.1.081
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 1, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 81-90
summary	Dataflow parametric modeling environments have become popular as exploratory tools due to them allowing the variational exploration of a design by controlling the parameters of its parametric model schema. However, the nature of these systems requires designers to prematurely commit to a structure and hierarchy of geometric relationships, which makes them inflexible when it comes to design exploration that requires topological changes to the parametric modeling graph. This paper is a first step towards augmenting parametric modeling systems via the use of machine learning for assisting the user towards topological exploration. In particular, this paper describes an approach where Long Short-Term Memory recurrent neural networks, trained on a data set of parametric modeling graphs, are used as generative systems for suggesting alternative dataflow graph paths to the parametric model under development.
keywords	design exploration; visual programming; machine learning
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:58

_id	ecaadesigradi2019_648
id	ecaadesigradi2019_648
authors	Eisenstadt, Viktor, Langenhan, Christoph and Althoff, Klaus-Dieter
year	2019
title	Generation of Floor Plan Variations with Convolutional Neural Networks and Case-based Reasoning - An approach for transformative adaptation of room configurations within a framework for support of early conceptual design phases
doi	https://doi.org/10.52842/conf.ecaade.2019.2.079
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 2, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 79-84
summary	We present an approach for computer-aided generation of different variations of floor plans during the early phases of conceptual design in architecture. The early design phases are mostly characterized by the processes of inspiration gaining and search for contextual help in order to improve the building design at hand. The generation method described in this work uses the novel as well as established artificial intelligence methods, namely, generative adversarial nets and case-based reasoning, for creation of possible evolutions of the current design based on the most similar previous designs. The main goal of this approach is to provide the designer with information on how the current floor plan can evolve over time in order to influence the direction of the design process. The work described in this paper is part of the methodology FLEA (Find, Learn, Explain, Adapt) whose task is to provide a holistic structure for support of the early conceptual phases in architecture. The approach is implemented as the adaptation component of the framework MetisCBR that is based on FLEA.
keywords	room configuration; adaptation; case-based reasoning; convolutional neural networks; conceptual design
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:55

_id	ecaadesigradi2019_605
id	ecaadesigradi2019_605
authors	Andrade Zandavali, Bárbara and Jiménez García, Manuel
year	2019
title	Automated Brick Pattern Generator for Robotic Assembly using Machine Learning and Images
doi	https://doi.org/10.52842/conf.ecaade.2019.3.217
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 3, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 217-226
summary	Brickwork is the oldest construction method still in use. Digital technologies, in turn, enabled new methods of representation and automation for bricklaying. While automation explored different approaches, representation was limited to declarative methods, as parametric filling algorithms. Alternatively, this work proposes a framework for automated brickwork using a machine learning model based on image-to-image translation (Conditional Generative Adversarial Networks). The framework consists of creating a dataset, training a model for each bond, and converting the output images into vectorial data for robotic assembly. Criteria such as: reaching wall boundary accuracy, avoidance of unsupported bricks, and brick's position accuracy were individually evaluated for each bond. The results demonstrate that the proposed framework fulfils boundary filling and respects overall bonding structural rules. Size accuracy demonstrated inferior performance for the scale tested. The association of this method with 'self-calibrating' robots could overcome this problem and be easily implemented for on-site.
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:54

_id	ijac201917102
id	ijac201917102
authors	Cutellic, Pierre
year	2019
title	Towards encoding shape features with visual event-related potential based brain–computer interface for generative design
source	International Journal of Architectural Computing vol. 17 - no. 1, 88-102
summary	This article will focus on abstracting and generalising a well-studied paradigm in visual, event-related potential based brain–computer interfaces, for the spelling of characters forming words, into the visually encoded discrimination of shape features forming design aggregates. After identifying typical technologies in neuroscience and neuropsychology of high interest for integrating fast cognitive responses into generative design and proposing the machine learning model of an ensemble of linear classifiers in order to tackle the challenging features that electroencephalography data carry, it will present experiments in encoding shape features for generative models by a mechanism of visual context updating and the computational implementation of vision as inverse graphics, to suggest that discriminative neural phenomena of event-related potentials such as P300 may be used in a visual articulation strategy for modelling in generative design.
keywords	Generative design, machine learning, brain–computer interface, design computing and cognition, integrated cognition, neurodesign, shape, form and geometry, design concepts and strategies
series	journal
email
last changed	2019/08/07 14:04

_id	ecaadesigradi2019_510
id	ecaadesigradi2019_510
authors	Giannopoulou, Effima, Baquero, Pablo, Warang, Angad, Orciuoli, Affonso and T. Estévez, Alberto
year	2019
title	Stripe Segmentation for Branching Shell Structures - A Data Set Development as a Learning Process for Fabrication Efficiency and Structural Performance
doi	https://doi.org/10.52842/conf.ecaade.2019.3.063
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 3, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 63-70
summary	This article explains the evolution towards the subject of digital fabrication of thin shell structures, searching for the computational design techniques which allow to implement biological pattern mechanisms for efficient fabrication procedures. The method produces data sets in order to analyse and evaluate parallel alternatives of branching topologies, segmentation patterns, material usage, weight and deflection values as a user learning process. The importance here is given to the selection of the appropriate attributes, referring to which specific geometric characteristics of the parametric model are affecting each other and with what impact. The outcomes are utilized to train an Artificial Neural Network to predict new building information based on new combinations of desired parameters so that the user can decide and adjust the design based on the new information.
keywords	Digital Fabrication; Shell Structures; Segmentation; Machine Learning; Branching Topologies; Bio-inspired
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:51

_id	ijac201917204
id	ijac201917204
authors	Karaoglan Füsun Cemre and Sema Alaçam
year	2019
title	Design of a post-disaster shelter through soft computing
source	International Journal of Architectural Computing vol. 17 - no. 2, 185-205
summary	Temporary shelters become a more critical subject of architectural design as the increasing number of natural disasters taking place each year result in a larger number of people in need of urgent sheltering. Therefore, this project focuses on designing a temporary living space that can respond to the needs of different post-disaster scenarios and form a modular system through differentiation of units. When designing temporary shelters, it is a necessity to deal with the provision of materials, low-cost production and the time limit in the emergency as well as the needs of the users and the experiential quality of the space. Although computational approaches might lead to much more efficient and resilient design solutions, they have been utilized in very few examples. For that reason and due to their suitability to work with architectural design problems, soft computing methods shape the core of the methodology of the study. Initially, a digital model is generated through a set of rules that define a growth algorithm. Then, Multi-Objective Genetic Algorithms alter this growth algorithm while evaluating different configurations through the objective functions constructed within a Fuzzy Neural Tree. The struggle to represent design goals in the form of Fuzzy Neural Tree holds potential for the further use of it for architectural design problems centred on resilience. Resilience in this context is defined as a measure of how agile a design is when dealing with a major sheltering need in a post-disaster environment. Different from the previous studies, this article aims to focus on the design of a temporary shelter that can respond to different user types and disaster scenarios through mass customization, using Fuzzy Neural Tree as a novel approach. While serving as a temporary space, the design outcomes are expected to create a more neighbourhood-like pattern with a stronger sense of community for the users compared to the previous examples.
keywords	Humanitarian design, emergency architecture, computational design, Fuzzy Neural Tree, Multi-Objective Genetic Algorithms
series	journal
email
last changed	2019/08/07 14:04

_id	caadria2019_298
id	caadria2019_298
authors	Karoji, Gen, Hotta, Kensuke, Hotta, Akito and Ikeda, Yasushi
year	2019
title	Pedestrian Dynamic Behaviour Modeling - An application to commercial environment using RNN framework
doi	https://doi.org/10.52842/conf.caadria.2019.1.281
source	M. Haeusler, M. A. Schnabel, T. Fukuda (eds.), Intelligent & Informed - Proceedings of the 24th CAADRIA Conference - Volume 1, Victoria University of Wellington, Wellington, New Zealand, 15-18 April 2019, pp. 281-290
summary	The research of developing and improving pedestrian simulation model is essential in the process of analysing, evaluating and generating the architectural spaces that can not only satisfy circulation design condition but also promote sales by attracting customers. In terms of programming the simulation for commercial environment, current study attempts to use shortest-path algorithm generally and these results suggested that the model can reproduce approximate real trajectory within given environment. However, these studies also mentioned about necessity of considering shopper internal state and visual field. In this paper, in order to further incorporate the dynamic internal state (memory) into simulation model, we propose using iterative algorithm based on recurrent neural network (RNN) framework which allow it to exhibit temporal dynamic behaviour for a time sequence. Finally, we demonstrate the effectiveness of these algorithms we introduce and assess the combination of multiple algorithms and calibration of probability by comparing with trajectories of the experiment.
keywords	Pedestrian simulation; Algorithm; RNN; Commercial environment
series	CAADRIA
email
last changed	2022/06/07 07:52

_id	caadria2019_449
id	caadria2019_449
authors	Lin, Yuqiong, Yao, Jiawei, Huang, Chenyu and Yuan, Philip F.
year	2019
title	The Future of Environmental Performance Architectural Design Based on Human-Computer Interaction - Prediction Generation Based on Physical Wind Tunnel and Neural Network Algorithms
doi	https://doi.org/10.52842/conf.caadria.2019.2.633
source	M. Haeusler, M. A. Schnabel, T. Fukuda (eds.), Intelligent & Informed - Proceedings of the 24th CAADRIA Conference - Volume 2, Victoria University of Wellington, Wellington, New Zealand, 15-18 April 2019, pp. 633-642
summary	As the medium of the environment, a building's environment performance-based generative design cannot be separated from intelligent data processing. Sustainable building design should seek an optimized form of environmental performance through a complete set of intelligent induction, autonomous analysis and feedback systems. This paper analyzed the trends in architectural design development in the era of algorithms and data and the status quo of building generative design based on environmental performance, as well as highlighting the importance of physical experiments. Furthermore, a design method for self-generating environmental performance of urban high-rise buildings by applying artificial intelligence neural network algorithms to a customized physical wind tunnel is proposed, which mainly includes a morphology parameter control and environmental data acquisition system, code translation of environmental evaluation rules and architecture of a neural network algorithm model. The design-oriented intelligent prediction can be generated directly from the target environmental requirements to the architectural forms.
keywords	Physical wind tunnel; neural network algorithms; dynamic model; environmental performance; building morphology self-generation
series	CAADRIA
email
last changed	2022/06/07 07:59

_id	caadria2020_259
id	caadria2020_259
authors	Rhee, Jinmo, Veloso, Pedro and Krishnamurti, Ramesh
year	2020
title	Integrating building footprint prediction and building massing - an experiment in Pittsburgh
doi	https://doi.org/10.52842/conf.caadria.2020.2.669
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 669-678
summary	We present a novel method for generating building geometry using deep learning techniques based on contextual geometry in urban context and explore its potential to support building massing. For contextual geometry, we opted to investigate the building footprint, a main interface between urban and architectural forms. For training, we collected GIS data of building footprints and geometries of parcels from Pittsburgh and created a large dataset of Diagrammatic Image Dataset (DID). We employed a modified version of a VGG neural network to model the relationship between (c) a diagrammatic image of a building parcel and context without the footprint, and (q) a quadrilateral representing the original footprint. The option for simple geometrical output enables direct integration with custom design workflows because it obviates image processing and increases training speed. After training the neural network with a curated dataset, we explore a generative workflow for building massing that integrates contextual and programmatic data. As trained model can suggest a contextual boundary for a new site, we used Massigner (Rhee and Chung 2019) to recommend massing alternatives based on the subtraction of voids inside the contextual boundary that satisfy design constraints and programmatic requirements. This new method suggests the potential that learning-based method can be an alternative of rule-based design methods to grasp the complex relationships between design elements.
keywords	Deep Learning; Prediction; Building Footprint; Massing; Generative Design
series	CAADRIA
email
last changed	2022/06/07 07:56

_id	acadia19_246
id	acadia19_246
authors	Zhang, Viola; Qian, William; Sabin, Jenny
year	2019
title	PolyBrickH2.0
doi	https://doi.org/10.52842/conf.acadia.2019.246
source	ACADIA 19:UBIQUITY AND AUTONOMY [Proceedings of the 39th Annual Conference of the Association for Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-59179-7] (The University of Texas at Austin School of Architecture, Austin, Texas 21-26 October, 2019) pp. 246-257
summary	This project emerged from collaborative trans-disciplinary research between architecture, engineering, biology, and materials science to generate novel applications in micro-scale 3D printed ceramics. Specifically, PolyBrick H2.0 adapts internal bone-based hydraulic networks through controlled water flow from 3D printed micro-textures and surface chemistry. Engagement across disciplines produced the PolyBrick series at the Sabin Lab (Sabin, Miller, and Cassab 2014) . The series is a manifestation of novel digital fabrication techniques, bioinspired design, materials inquiry, and contemporary evolutions of building materials. A new purpose for the brick is explored that is not solely focused on the mechanical constraints necessary for built masonry structures. PolyBrick H2.0 interweaves the intricacies of living systems (beings and environments combined) to create a more responsive and interactive material system. The PolyBrick 2.0 series looks at human bone as a design model for foundational research. PolyBrick H2.0 merges the cortical bone hydraulic network with new functionalities as a water filtration and collection system for self-preservation and conservation as well as passive cooling solutions. It also pushes the ability of 3D printing techniques to the microscale. These functionalities are investigated under context for a better construction material, but its use may extend further.
series	ACADIA
type	normal paper
email
last changed	2022/06/07 07:57

_id	ecaadesigradi2019_034
id	ecaadesigradi2019_034
authors	Chen, Dechen, Luo, Dan, Xu, Weiguo, Luo, Chen, Shen, Liren, Yan, Xia and Wang, Tianjun
year	2019
title	Re-perceive 3D printing with Artificial Intelligence
doi	https://doi.org/10.52842/conf.ecaade.2019.1.443
source	Sousa, JP, Xavier, JP and Castro Henriques, G (eds.), Architecture in the Age of the 4th Industrial Revolution - Proceedings of the 37th eCAADe and 23rd SIGraDi Conference - Volume 1, University of Porto, Porto, Portugal, 11-13 September 2019, pp. 443-450
summary	How can machine learning be combined with intelligent construction, material testing and other related topics to develop a new method of fabrication? This paper presents a set of experiments on the dynamic control of the heat deflection of thermoplastics in searching for a new 3D printing method with the dynamic behaviour of PLA and with a comprehensive workflow utilizing mechanic automation, computer vision, and artificial intelligence. Additionally, this paper will discuss in-depth the performance of different types of neural networks used in the research and conclude with solid data on the potential connection between the structure of neural networks and the dynamic, complex material performance we are attempting to capture.
keywords	3D printing; AI; automation; material; fabrication
series	eCAADeSIGraDi
email
last changed	2022/06/07 07:55

_id	acadia19_412
id	acadia19_412
authors	Del Campo, Matias; Manninger, Sandra; Carlson, Alexandra
year	2019
title	Imaginary Plans
doi	https://doi.org/10.52842/conf.acadia.2019.412
source	ACADIA 19:UBIQUITY AND AUTONOMY [Proceedings of the 39th Annual Conference of the Association for Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-59179-7] (The University of Texas at Austin School of Architecture, Austin, Texas 21-26 October, 2019) pp. 412-418
summary	Artificial Neural Networks (NN) have become ubiquitous across disciplines due to their high performance in modeling the real world to execute complex tasks in the wild. This paper presents a computational design approach that uses the internal representations of deep vision neural networks to generate and transfer stylistic form edits to both 2D floor plans and building sections. The main aim of this paper is to demonstrate and interrogate a design technique based on deep learning. The discussion includes aspects of machine learning, 2D to 2D style transfers, and generative adversarial processes. The paper examines the meaning of agency in a world where decision making processes are defined by human/machine collaborations (Figure 1), and their relationship to aspects of a Posthuman design ecology. Taking cues from the language used by experts in AI, such as Hallucinations, Dreaming, Style Transfer, and Vision, the paper strives to clarify the position and role of Artificial Intelligence in the discipline of Architecture.
series	ACADIA
type	normal paper
email
last changed	2022/06/07 07:55

_id	cf2019_005
id	cf2019_005
authors	Eisenstadt, Viktor; Klaus-Dieter Althoff and Christoph Langenhan
year	2019
title	Supporting Architectural Design Process with FLEA A Distributed AI Methodology for Retrieval, Suggestion, Adaptation, and Explanation of Room Conﬁgurations
source	Ji-Hyun Lee (Eds.) "Hello, Culture!" [18th International Conference, CAAD Futures 2019, Proceedings / ISBN 978-89-89453-05-5] Daejeon, Korea, p. 24
summary	The artiﬁcial intelligence methods, such as case-based reasoning and artiﬁcial neural networks were already applied to the task of architectural design support in a multitude of speciﬁc approaches and tools. However, modern AI trends, such as Explainable AI (XAI), and additional features, such as providing contextual suggestions for the next step of the design process, were rarely considered an integral part of these approaches or simply not available. In this paper, we present an application of a distributed AI-based methodology FLEA (Find, Learn, Explain, Adapt) to the task of room conﬁguration during the early conceptual phases of architectural design. The implementation of the methodology in the framework MetisCBR applies CBR-based methods for retrieval of similar ﬂoor plans to suggest possibly inspirational designs and to explain the returned results with speciﬁc explanation patterns. Furthermore, it makes use of a farm of recurrent neural networks to suggest contextually suitable next conﬁguration steps and to present design variations that show how the designs may evolve in the future. The ﬂexibility of FLEA allows for variational use of its components in order to activate the currently required modules only. The methodology was initialized during the basic research project Metis (funded by German Research Foundation) during which the architectural semantic search patterns and a family of corresponding ﬂoor plan representations were developed. FLEA uses these patterns and representations as the base for its semantic search, explanation, next step suggestion, and adaptation components. The methodology implementation was iteratively tested during quantitative evaluations and user studies with multiple ﬂoor plan datasets.
keywords	Room con?guration, Distributed AI, Case-based reasoning, Neural networks, Explainable AI
series	CAAD Futures
type	normal paper
email
last changed	2019/07/29 14:11

_id	cf2019_010
id	cf2019_010
authors	Lorenz, Clara-Larissa; Bleil De Souza, Spaeth and Packianather
year	2019
title	Machine Learning in Design Exploration: An Investigation of the Sensitivities of ANN-based Daylight Predictions
source	Ji-Hyun Lee (Eds.) "Hello, Culture!" [18th International Conference, CAAD Futures 2019, Proceedings / ISBN 978-89-89453-05-5] Daejeon, Korea, pp. 75-87
summary	The use of Artificial Neural Networks (ANNs) promises greater efficiency in the assessment of daylight situations than simulations. With the daylight factor under scrutiny and the recent adaptation of climate-based daylight metrics in British and European buildings standards, ANNs provide a possibility for instantaneous feedback on otherwise time-consuming performance metrices. This study demonstrates the application of ANNs as prediction systems in design exploration. A specific focus of the research is the flexibility of ANNs, their reliability and sensitivity to changes.
keywords	Artificial neural networks, atria, climate-based daylight modeling, daylight autonomy, daylight performance, parametric design
series	CAAD Futures
email
last changed	2019/07/29 14:08

For more results click below: