21793 research outputs found
Sort by
Generalization of Urban Wind Field Using Fourier Neural Operators Across Different Wind Directions and Cities
Get PDFIn urban environments, the most common forms of air transportation are helicopters and
unmanned aerial vehicles (UAVs). There is a high demand for air transport of small and medium
sized aircraft, including UAVs. Wind field simulations in urban environments are typically
performed using computational fluid dynamics (CFD), and most of these models fall into the
categories of direct numerical simulation (DNS) and large eddy simulation (LES). Although
these models are accurate, they are time-consuming, so there is a need to develop a more
convenient method to replace the traditional CFD methods. In recent years, with the rapid
development of artificial intelligence technology and graphics processing unit (GPU) hardware, a
promising research direction has emerged. Currently, many studies are using artificial
intelligence-based deep learning techniques to transform the computational processes associated
with wind field simulation. The goal of these studies is not only to achieve the accuracy of
traditional CFD models, but to surpass them while significantly accelerating the computational
process. In this paper, we apply the Fourier Neural Operator (FNO) method based on deep
learning technology to simulate the wind field in a two-dimensional urban environment. The
method uses a Fourier module to extract and learn features in the Fourier frequency domain of
the input data. Compared to traditional convolutional neural network (CNN) modules, Fourier
modules aim to learn global features in the Fourier frequency domain of the input data. In
contrast, a convolutional neural network (CNN) module performs feature learning in the local
spatial domain of the input features. In addition, the input features are processed by a Multi-
Layer Perceptron (MLP) module, and the feature output of the MLP module is added to the
feature output of the Fourier module. This structure is based on a residual network (ResNet),
which can mitigate the phenomenon of gradient vanishing or gradient explosion that occurs
when input data propagates through a multilayer network.
The FNO model ultimately maps the input features (i.e., the input wind field) to the desired
output features (i.e., the output wind field dimensions). Gradients are updated through back
propagation to reduce the discrepancy between the FNO model’s output wind field and the actual
wind field, thus facilitating the deep learning process. After a series of experiments, the optimal
settings for the Fourier layer number and the intermediate feature dimensions of the MLP in the
FNO model were determined. In this context, “intermediate feature dimensions” refers to the
number of features extracted by the MLP module. These settings ensure that the FNO model
achieves the best results on the dataset while minimizing computational overhead and resource
consumption. The training phase utilized wind field data from Niigata with westerly winds, with
a time step of 0.1 seconds, and the output consisted of wind fields at the same location with a
time step of 1 second (i.e. 10 time steps). Experimental results demonstrated that the FNO model
could predict the wind field over the entire Niigata urban area for the next 7 seconds (i.e. 70 time
i
steps), with an average absolute error of less than 0.5 m/s. Importantly, the FNO exhibited strong
generalization capabilities in different wind conditions: although the training data consisted of
westerly wind data from Niigata, the model performed well in tests with northerly winds. Further
validation across different urban geometries revealed that the FNO model could accurately
predict 70 time steps (7 seconds) of wind fields in the vertically flipped version of Niigata,
indicating that it generalizes well when the geometry is similar to the training data. However, in
Montreal, which has a significantly different urban geometry, the model’s accuracy diminished
after 10 time steps. This highlights the significance of urban geometry in wind field prediction.
During this process, the FNO’s wind field simulation was 300 times faster than that of the
CityFFD model we employed, with CityFFD requiring 2.2 seconds per step, whereas FNO took
only 0.006 seconds. This further underscores the potential of the FNO model for practical
applications in wind field simulation.
Although it is premature to use FNO directly to replace wind field simulation due to the
exponential growth of errors with time, it is possible to use it in conjunction with CityFFD and
other technologies as a complementary model. For example, the wind field output by CityFFD at
a given time step can be used as input to FNO, which can generate the wind field in the same
area at subsequent time steps. The final output wind field can be used as input to CityFFD, thus
reducing the intermediate computation time of CityFFD
BIM-Based Automated Fault Detection and Diagnosis of HVAC Systems Using Knowledge Models
Get PDFAutomated Fault Detection and Diagnosis (AFDD) of building mechanical systems, including HVAC
(Heating, Ventilation, and Air Conditioning), has received substantial attention recently from both research
and application angles. The reasons are attributed to potential savings in energy consumption and
maintenance. Various methods, including simulation and Grey-Box, are offered, but data-driven ones have
received the most attention due to reduced manual effort, integrability, and scalability. Accordingly, to
enhance energy efficiency and reduce operational costs, various Machine Learning (ML) models have
been developed for AFDD of HVAC systems. However, the implementation of such data-driven
approaches has often translated into a loss of contextual data. This study integrates operational data with
building information and its various disciplines, linking the two to facilitate AFDD model development. BIM
(Building Information Model) and BAS/BMS (Building Automation System/Building Management System)
data are the repositories utilized for this integration.
The proposed solution integrates bottom-up (data-driven via Machine Learning) and top-down
(knowledge-oriented via Semantic Web Technologies) AI approaches to generate an effective AFDD
knowledge model. The study materializes a two-way flow of data and knowledge between the BIM and
BMS by utilizing an ontology named AFDDOnto, which integrates building components with fault types,
methods, and parameters. The solution enables AFDD algorithms to utilize static and dynamic information
related to HVAC and building spaces to develop enriched AFDD models. It incorporates building spatial
information and stores analytics to represent the facility's as-is state. The proposed BIM-based knowledge
solution can be used for AFDD model development, tracking changes, and analysis and visualization in
two ways. Firstly, to integrate the BIM features with BMS features for creating ‘context-aware’ AFDD
models. Secondly, to semantically store BIM-based AFDD performance analytics through AFDDOnto that
can be used for model comparison, reproduction and visualization through knowledge graphs.
The knowledge stored in the repository can be queried, which enables access to contextual information
(knowledge graphs, images, videos, project snippets); spatial data (locations, states); and apriori
knowledge (configuration and analytics) to enable development, application, and visualization of context-
aware AFDD models. Additionally, the proposed solution can maintain access to external project files and
databases to enable interoperability between BIM and BAS/BMS. The potential users include HVAC
operators, BIM Managers, and Facility Managers tasked with the operation and maintenance of HVAC
systems
The Making of the Management Machine: Subjectivity, Desire, and Living Labour
No full textThis thesis explores how managerial practices and workers operate within capitalist work relations, focusing on how they influence and are influenced by the production of subjectivities across various historical and social contexts. Central to this study is the concept of living labour—the dynamic and creative capacity of workers to generate value through their labour. In capitalist economies, living labour is managed through strategies that aim to maximize productivity and efficiency, often involving control, coordination, and standardization. Consequently, management’s role is to organize and regulate living labour in alignment with the interests of capital, converting workers' potential into measurable outputs. This process reflects a core principle of capitalism: that labour must be continually transformed and subsumed by capital to generate profit.
Drawing on Deleuze and Guattari’s concept of desire as a productive force that generates connections and drives creation between people, objects, and systems, I argue that managerial practices play a critical role in shaping the ongoing formation of new configurations and possibilities. By theorizing the production of subjectivity as an open process continually actualized in the social realm, this project uncovers how social relations—whether in early industrial factories or contemporary tech offices—become terrains of struggle. Through interviews with managerial workers in tech startups, this research examines the affective and immaterial dimensions of managerial labour, showing how it functions as a vital component in securing the conditions necessary for capital accumulation
Frames for measures on Cantor sets
Get PDFIn this study of frames for measures on Cantor sets, we consider four measures with support contained in a Cantor set. These are the mass distribution measure, the Hausdorff measure of the appropriate dimension restricted to the Cantor set, the unique measure from Hutchinson's theorem for self-similar sets and the unique Lebesgue-Stieltjes measure with respect to the Cantor-Lebesgue function. For the ternary and quaternary Cantor sets, respectively, we show that
these four measures give the same measure . This allows us to study frames in .
While the theory of frames is well developed, the literature on frames on Cantor
sets is recent and limited. Central in defining frames of exponentials on Cantor sets is the set of integers (hereafter called \textit{spectrum}) obtained from the Fourier transform of each measure supported on the corresponding Cantor set. After giving some background on frames, we follow the work of Jorgensen and
Pedersen (1998) to find the spectrum of the mass distribution measure on the quaternary Cantor set from its Fourier transform and show that we do have an orthonormal basis, which is a special case of a frame. We also present the result of Jorgensen and Pedersen (1998) for the mass distribution measure on the ternary Cantor set, that it is not possible to have a spectrum that yields an orthonormal set of exponentials. However, this leads to the question: can we show the existence of a frame from the spectrum of the mass distribution measure on the ternary Cantor set? Recent work (for example, Lev (2018), Picioroaga and Weber (2017)) study this question but it remains an open problem
Lightweight Authentication for Edge Computing
Get PDFEdge computing (EC) is one of the most promising decentralized network paradigms in the proliferation era of the Internet of Things (IoT). Although this paradigm has high potential in terms of performance and de-centralization, it carries several security concerns. One of the most important security properties for the EC paradigm is authenticity. It allows different edge computing entities to verify each other through cryptographic means. Additionally, authenticity regulates access control to different edge computing resources and data. There are several types of authentications, one-way authentication, mutual authentication, broadcast authentication, group authentication, and others. In our thesis, we focus on designing security protocols for different edge computing applications that require either mutual authentication, broadcast authentication, or group authentication. In all our proposed protocols we utilize lightweight security primitives suitable for the three-tier cloud-edge-IoT architecture.
In the first protocol, we utilize lightweight cryptographic primitives to design mutual authentication for EC broadcast messages. Specifically, the used primitives are only hash-based one-way chain, symmetric-key cryptography, and a hash function. The protocol establishes key-agreement for group and individual nodes in each session. The achieved security properties for our protocol are mutual-authentication for broadcast messages, message secrecy, message integrity, and forward secrecy. We formally define and prove the main security properties of our protocol theoretically using the indistinguishability game. We compare our protocol to other lightweight protocols in terms of security and performance to prove its advantages in terms of computations, communication overhead, and storage.
Motivated by the fact that mass authentication is one of the desirable security features in the edge computing paradigm, our second proposed protocol is a lightweight group authentication scheme (GAS) with session key-agreement. The protocol utilizes lightweight cryptographic primitives, namely, Shamir’s secret sharing (SSS) scheme and aggregated message authentication code (Aggregated-MAC). Unlike other group authentication schemes, our protocol provides multiple asynchronous authentications. Furthermore, we implement a simple key refreshing mechanism such that in each session, a new session-key between group nodes and the authenticating server is established without the need for redistributing new shares. Our security analysis includes proving that our protocol provides group authenticity, message forward secrecy, and prevents several attacks.
Extending our group authentication design, our third security protocol is a flexible GAS based on Physical Unclonable Function (PUF) and Shamir’s secret sharing scheme. Specifically, we apply PUFs on SSS and utilize the SSS-homomorphic property to achieve multiple-time group authentications with the same set of shares. Our scheme is lightweight, establishes a new group key-agreement per session, and supports efficient node-evicting mechanism. Furthermore, in our protocol, the group nodes do not store any shares; instead, the nodes derive their secret-shares from their PUF-responses. We formally analyze our protocol theoretically and with automated tools, Automated Validation of Internet Security Protocol and Applications (AVISPA), to prove that our scheme achieves message secrecy and authenticity.
Finally, we propose a lightweight symmetric-key based protocol which provides edge-IoT mutual authentication, forward secrecy, backward secrecy, and anonymity. The security primitives used in our fourth protocol are pseudo-random function (PRF), random number generation, a hash function, and xor. We prove the security goals of the protocol and compare it to other lightweight authentication protocols
Interrogating The Concurrence Of Graffiti And The Built Environment For Future Urban Renewal.
Get PDFThe main purpose of this paper is to examine the phenomenon of graffiti in the urban environment in relation to urban design and architectural decisions. This research creation hopes to shed light on areas of the city that require a more insightful evaluation prior to the consideration and adaptation as pre design research for future urban design projects. Post occupancy evaluation at the scope of a city can benefit from a new metric, one that has always been dismissed as irrelevant yet may have a trend that can help future evaluations or propositions of urban proposals. An emphasis on transdisciplinary research using autographic visualization to extract intersubjective data seeks to ameliorate future projects. Using a phenomenological inquiry perspective to navigate the results of qualitative methodologies results in recommendations for the future of our cities
Functional Studies of Protein Assemblies Involved in Bacterial Siderophore Biosynthesis
No full textIron is an indispensable micronutrient for most microorganisms. A high-affinity iron chelator called enterobactin is synthesized by Escherichia coli to scavenge scarce ferric iron from its extracellular environment. The biosynthesis of enterobactin is carried out by sequential orchestration of six enzymes, EntC, EntB, EntA, EntE, EntF, and EntD. Previous studies on enterobactin biosynthesis and other siderophores, like pyoverdine and bacillibactin, indicate protein-protein interactions (PPIs) are extensively engaged in these processes, leading to the thesis research goals of ongoing mapping and functional characterization of this PPI. We hypothesize these interactions, albeit weak, may occur among EntC, EntB, and EntA which are responsible for synthesizing 2,3-dihydroxybenzoate (2,3-DHB), a precursor for enterobactin. Furthermore, we hypothesize that a higher-order enzymatic assembly (i.e., a metabolon) forms in the E. coli cytoplasm to facilitate efficient catalysis and regulation of iron homeostasis at the protein level. Here the first evidence of an interaction between EntC and EntB both in vivo and in vitro is reported, thereby completing the map of pairwise PPIs among all the enterobactin biosynthetic enzymes. Further characterization reveals an electropositive channeling surface between the EntC active site and EntB isochorismate active site, suggesting an electrostatic channeling mechanism of enhancing metabolic flux by preventing labile intermediates from diffusing into the bulk solution. Given EntB’s central role in enterobactin biosynthesis, this enzyme is likely a target for regulation. A novel role for 2,3-DHB as a feedback inhibitor of EntB isochorismatase activity was identified, as well as its effect on the down-regulation to the EntCB isochorismate channeling. Finally, a chimeric protein was constructed, composed of dimeric and enzymatically active EntB and EntA fused components, paving the way to investigate the potential existence of higher-order Ent complexes and their recognition of cognate partners
Queer City Film: Experimental Cinema, Urban Life and the Erotics of Visibility
No full textThe “queer city film” is proposed in this dissertation to be a corpus of experimental cinema and a critical method of spectatorship that establishes links between queer film history and studies of the cinematic city, two fields that have developed in parallel with little explicit overlap. The city is thus analyzed as an opportune vantage point for the discussion of concurrent inventions of homosexuality and cinema in Euro-American modernity. By establishing a continuity between the cinema-and-the-city discourse and queer film theory, this project identifies the emergence of subjective vision in modernity as a precondition for queer vision, thus proposing a new interpretative model for queer cinema that captures historical mediations of divergent expressions of gender and sexuality in visual media. Focusing on queer-authored experimental cinema set in recognizable urban locales across North America and Europe, the project puts film analysis into dialogue with queer theory, material culture studies, fashion and design studies, and cultural geography. To that end, city symphonies are analyzed as the earliest examples of the cinematic city representing a hermeneutics for the visual development of gender and sexual spectrums throughout film history. Gay pornography produced in the United States and Quebec is examined as a genre that put eroticism into dialogue with a documentary impulse, capturing the politically charged issues of queer visibility and public infrastructure through on-location shooting in cities such as San Francisco and Montreal. Finally, pre-digital queer experimental cinema of the twentieth century is surveyed for its sustained, yet often overlooked, interest in the closet of sexual secrecy as a conceptual and physical space denoting a range of domestic practices. By privileging experimental, avant-garde and erotic cinema over commercial film, and by gravitating towards the mediations of urban figures, spaces and surfaces over the narrative plot in queer cinema, the dissertation proposes an intermedial and historical approach to cinema as an archive of queer meanings and fantasies that challenges neat categories of gender and sexual identity, and expands their representational possibilities
Genetic network rewiring between distantly related eukaryotic species
No full textGenetic interactions occur when a combination of mutations in different genes results in an unexpected phenotype based on the combined effects of individual single mutants. Synthetic lethality represents an extreme example of a genetic interaction resulting in lethal double mutant from the combination of viable single mutants. The knowledge of the genetic interactions that underlie conditional gene essentiality across species, whereby a gene which is required for viability in one species is dispensable in another has remained elusive. In this thesis, I investigated digenic interactions that underlie conditional essentiality, and the trigenic interactions that underlie conditional synthetic lethality in distantly related eukaryotes to understand the principles of genetic network conservation. First, I used literature curation to identify conditional essential genes and their synthetic lethal interactions across S. cerevisiae, S. pombe, C. elegans and H. sapiens. I applied functional enrichment analysis and demonstrated that conditional essential genes are rewired by functionally related and specific synthetic lethal digenic interactions. Second, I identified synthetic lethal double mutants in S. cerevisiae that were viable in S. pombe. I constructed and screened 14 conditional essential double mutants and corresponding 24 single mutant query strains using trigenic Synthetic Genetic Array (τ-SGA) to identify trigenic interactions. I found that similar to single mutants, conditional synthetic lethal double mutants were rewired by functionally related trigenic interactions. Ultimately, understanding the rewiring of gene essentiality by genetic interactions sheds light on genome evolution and synthetic lethal therapies for human disease
Advancing Short-term Bus Passenger Flow Prediction with Graph Neural Network Models
No full textPredicting short-term passenger flow in urban bus networks is a crucial task for optimizing transit operations, reducing congestion, and enhancing transit commuter experience. This thesis introduces several innovative deep learning models aimed at addressing the unique challenges of bus networks, including temporal dynamics, spatial variability, and the impact of real-time traffic conditions. To model the complex relationships in transit networks, we leverage Graph Neural Networks (GNNs), which are particularly well-suited for capturing the non-Euclidean structure of bus networks.
In the first model, a Bus Network Graph Convolutional Long Short-Term Memory (BNG-ConvLSTM) neural network is developed to forecast short-term passenger flow. This model outperforms traditional deep learning models in scalability and robustness, as validated by real-world data from the Laval bus network. Extending this, we introduce the Traffic-Aware Multistep Graph Neural Network (TMS-GNN), which integrates traffic conditions and addresses the issue of exposure bias in multistep forecasting by employing Scheduled Sampling. This model significantly improves accuracy in multistep prediction and better adapts to the realities of urban traffic patterns.
To further capture the dynamic nature of public transportation, we propose Spatial-Temporal Attention Masked Graph Encoder-Decoder (STAM-GED), which integrates real-time bus schedules to model both node and edge changes in a network. This approach provides a more accurate representation of passenger flow, reflecting the real-time operational state of bus stops.
Finally, we explore transfer learning as a solution to the challenge of data scarcity, a common issue in many cities. We develop a transfer learning framework for GNNs, which uses a novel reinforcement learning optimization-based graph partitioning method to adapt models trained on data-rich networks to cities with limited data. This framework enables the transfer of knowledge across diverse urban environments, ensuring scalability and generalizability without sacrificing predictive accuracy.
Through comprehensive experiments on real-world data from multiple cities, including Ames-USA and Laval-Canada, our models demonstrate significant improvements in passenger flow prediction over existing methods. These contributions offer solutions for enhancing the reliability and efficiency of public transportation systems, paving the way for smarter, more sustainable urban mobility