26400 research outputs found
Sort by
I Often Dream of Trains
No full textThis is a collection of stories exploring travel, romantic illusions, motherhood and fairy-tale motifs through a surreal lens
Decoding Anisotropic Porous Medium: A Synergy Of Lattice Boltzmann Modelling And Operator Learning To Predict Permeability As A Function Of Orientation
No full textUnderstanding the directional properties of porous media is essential for accurately predicting flow behavior, reactive transport, and fluid-solid interactions in systems ranging from geothermal reservoirs to energy storage devices and biological tissues. Directional variations in permeability - reflecting a medium\u27s response to flow at different angular orientations - are particularly important for complex, inherently anisotropic geometries. In this study, we employ a Lattice Boltzmann (LBM) model to calculate directional permeabilities from porous media images subjected to varying flow inlet angles. Three classes of porous media were investigated: (1) synthetic media with circular grains, serving as isotropic baselines; (2) synthetic media with elliptical grains to introduce controlled anisotropy; and (3) micro-CT images of sandstone, characterized by naturally irregular grain distributions. For the synthetic cases, key geometric parameters were varied to diversify the medium structures. Using our LBM flow simulations, permeability was evaluated at 10° intervals across 360°, yielding 36 data points per sample. This systematic approach produced a comprehensive dataset capturing unique functional relationships between flow angle and permeability for each media class. Our primary goal is to analyze the anisotropy of these porous structures. While linear transformations of principal permeabilities can predict directional permeability in isotropic media, their applicability to anisotropic cases remains underexplored. We further aim to train machine learning models to predict permeability as a function of inlet flow angle given an image of the porous medium, and to compare model performance across the three classes of obstruction patterns. These findings deepen our understanding of how geometric anisotropy influences directional transport in porous media and pave the way for more accurate predictive models. Such advances have broad implications for filtration, energy storage, and subsurface fluid flow applications
Generative AI for 3D Printed Antenna Design
No full textThis research explores the integration of generative artificial intelligence (AI) with a physics-informed particle swarm optimizer (PSO) to develop 3D printable microstrip patch antennas. A neural network was trained on a dataset of microstrip patch antenna geometries and their corresponding performance metrics: return loss and gain. The PSO used a fitness function prioritizing low return loss in potential antennas, eventually yielding novel antenna geometries with parasitic components. 3D printing constraints were also hard coded into the framework, thus preventing any geometries being generated that cannot be fabricated. When simulated using Ansys HFSS, the AI generated microstrip patch antennas exceeded the target performance metrics, as the designed antenna demonstrated a gain increase of 0.75 dB above the target, and an S11 improvement of 6.48 dB beyond the target of -20 dB. These results illustrate the utility of AI with an optimization algorithm in generating antenna geometries that not only perform well, but are readily fabricable using hybrid 3D printing. The use of hybrid 3D printing in this area reduces the limitations found in traditional manufacturing and allows for greater exploitation of the degrees of freedom offered by the additive manufacturing technology. This will allow for rapid prototyping and a more explorative examination of the 3D design space in the future
Markerless clinical gait identification using computer vision techniques and artificial intelligence
No full textPose estimation algorithms could provide an unbiased examination for diagnosis, progression, and rehabilitation. PURPOSE: The purpose of this study was 1) to critically identify and appraise literature investigating spatiotemporal and kinematic gait characteristics in individuals with idiopathic Parkinson\u27s Disease and spastic diplegic Cerebral Palsy compared to healthy controls; 2) to examine the validity and reliability of spatiotemporal gait characteristics and upper- and lower-body joint kinematics of OpenPose and the agreement with marker-based motion capture in healthy individuals; and 3) to examine the sensitivity, specificity and predictive values of a proposed clinical diagnostic tool using OpenPose in the diagnosis of idiopathic Parkinson\u27s disease, spastic diplegic Cerebral Palsy and non-pathological gait. METHODS: First, meta-analysis was performed using the random-effects model for variables with at least three studies. Next, data (motion capture and digital video) were retrieved from an available dataset (Kwolek et al., 2019) and spatiotemporal gait characteristics and joint angle kinematics were calculated. One-way repeated measure ANOVAs, mean absolute errors, and ICC values were performed and calculated. Finally, reference data was retrieved from available literature, and 143 input videos were input into an identification MATLAB script. Sensitivity, specificity, positive predictive and negative predictive values, and likelihood ratios were calculated. RESULTS: Individuals with idiopathic Parkinson\u27s disease and spastic diplegic Cerebral Palsy exhibited gait deviations when compared to healthy individuals. OpenPose can estimate spatiotemporal gait characteristics and sagittal upper and lower extremity joint angles in healthy gait. OpenPose performed with 32.47% and 93.10% sensitivity for Parkinson\u27s disease and Cerebral Palsy, respectively. CONCLUSION: OpenPose could possess the capability to be implemented in a proposed diagnostic tool for distinguishing gait disorder
Interrogating Representational Rhetoric of Aid Discourse: Artifacts Analysis and Pedagogical Implications
No full textThis dissertation examines the representational strategies used by six aid organizations operating in Nepal, focusing on how they discursively construct socio-economic realities and power relations between themselves and their stakeholders. Employing a methodological framework that integrates Critical Discourse Analysis (CDA) and visual analysis within a broader content analysis approach, the study critically assesses website artifacts from Save the Children, Plan International, Oxfam, Care Nepal, ActionAid, and Christian Aid, along with insights from the writing instructors who participated in the study. Grounded in anti-essentialist theoretical perspectives drawn from Michel Foucault, Edward Said, Antonio Gramsci, and Stuart Hall, the research reveals that the aid discourse constructed by these organizations centers on influencing and persuading Western donors through hegemonic narratives of scarcity and vulnerability-narratives that often undermine the dignity and privacy of the stakeholders. Such representational rhetoric promotes the stakeholders\u27 dependency on aid organizations, reproduces existing inequalities, and obscures the strengths of local agencies, thereby legitimizing the role of international aid as an essential solution. By amplifying stakeholders\u27 problems and offering simplistic solutions, these narratives sustain global power asymmetries and perpetuate the West\u27s paternalistic ideology. Extending this decolonial rhetorical analysis to writing pedagogy, the dissertation proposes incorporating representation-related visual artifacts as a critical tool for enhancing students\u27 rhetorical analysis skills and developing their critical awareness of representation, identity, power, and discourse in writing classrooms. This study contributes to scholarship on representational rhetoric, visual rhetoric, and writing pedagogy in the context of Global South-Global North relations
You\u27re Softer Than the Western World
No full textA hybrid collection of stories and poems exploring the limits of identity and the possibility of solidarity
Effective Transformer Networks For Undersampled Magnetic Resonance Image Reconstruction
No full textThe proliferation of data-driven tools for solving problems in every possible domain, coupled with rapid advances in computing technology, has led to an arms race of AI development and application research in industry and academia. One field of research that stands to gain immeasurably from this revolution is medical imaging. It is a critical part of modern diagnostics, and advancements in this area can directly benefit the average person by making healthcare more accessible, accurate, and affordable. Breakthroughs in mainstream image processing and computer vision have long fueled development in medical imaging, and it is now common to see cutting edge machine vision models developed less than a year ago being explored to solve problems in this field. Convolutional models are already being tested in commercial scanners for a variety of tasks such as pathology detection, organ segmentation, and image denoising, with Vision Transformers lining up to be the next big breakthrough.
This Ph.D. research focuses on the effective use of the notoriously large and unwieldy Transformer models to accelerate Magnetic Resonance Imaging (MRI) scans. To this end, it explores the use of deep neural networks for undersampled multi-channel reconstruction of complex MRI data in a k-space data consistent fashion with a focus on Shifted-Window (Swin) Transformers. Specifically, cascaded reconstruction pipelines were designed using customized Swin Transformer blocks featuring overlapped window attention, and their performance tested against the state-of-the-art methods.
The results show that cleverly designed small Transformer architectures working together can outperform larger monolithic structures. Additionally, an exploration of transfer learning in this area revealed that it is possible to leverage the abundance of widely available magnitude MR images to pre-train transformer blocks and adapt them into cascaded reconstruction architectures to be fine-tuned for multi-channel complex reconstruction tasks
Enhancing Security and Resiliency in Operational Technology Environments Through Network Slicing and Federated Learning
No full textThe growing convergence of Information Technology (IT) and Operational Technology (OT) within Industry 4.0 environments has introduced new demands on industrial network infrastructure. As cyber-physical systems become increasingly interconnected, ensuring the secure, timely, and efficient exchange of critical data is essential. This thesis explores how network slicing, a method of creating isolated virtual network segments, can be applied within OT environments to address challenges such as latency, security, and resource allocation.
The first research question addressed in this thesis is: How can OT networks take advantage of NFV and SDN technology to become cyber resilient? This study examines the operational, security, and architectural implications of introducing network slicing into traditionally static OT infrastructures such as Industrial Control Systems (ICS) and SCADA. Through simulated deployments and case studies, the research demonstrates how slicing enables better isolation between critical and non-critical services, thereby improving response time, throughput, and security in sensitive environments.
The second question considers: How to dynamically implement network slicing and take advantage of network resources towards integrating decentralized machine learning? In response, this thesis proposes a framework that combines Software-Defined Networking (SDN), Network Function Virtualization (NFV), and Federated Learning (FL) to enable real-time analytics while maintaining data locality. The proposed approach reduces the burden on centralized infrastructure and minimizes privacy risks by supporting on-site training of models across distributed OT nodes, coordinated through dynamically allocated network slices.
The third focus explores: How slicing helps to increase the resiliency of OT networks through the orchestration of a dynamic DMZ? To answer this, the thesis presents a method for creating and managing Dynamic Demilitarized Zones (DMZs) using network slicing. This enables flexible and automated isolation of sensitive subsystems during threat scenarios or high-risk operations. Coupled with intelligent orchestration and containerized security services, the dynamic DMZ significantly enhances the system\u27s ability to respond to cyber incidents without halting production.
Ultimately, this thesis contributes a comprehensive architecture that blends network slicing with machine learning, secure segmentation, and automation, paving the way for resilient, adaptive, and intelligent OT environments. Performance evaluations across multiple scenarios show improvements in system reliability, threat response time, model accuracy, and resource utilization, providing a strong foundation for future industrial automation systems
H.O.V. - A Novel
No full textIn this novel. An agribusiness executive from Toronto lives as a homeless man in 1990\u27s Los Angeles. He\u27s wanted for a crime in Canada and is unable to find conventional work. His living involves being a professional passenger for morning and afternoon commuters who wish to use the High Occupancy Vehicle lane. After falling in love with a clerk at a used paperback store, he must confront his choices and deal with the circumstances of his wife\u27s death