22505 research outputs found
Sort by
NestView-Transforming house hunting with AR
Get PDFNestView is an augmented reality (AR) based smart glasses system which aims to facilitate the process of finding rental homes for people. Conventional rental websites bombard users with two-dimensional data and have no sense of place, which is why they\u27re both time-consuming and mentally taxing. NestView gives the user real-time mobile listings displayed in AR right in their physical space, providing an immersive, contextually aware rental experience. Informed by a user-centered design methodology, this study examines what frustrations the renters experience, and in what ways AR applications can meet those frustrations. The system comes with proximity-driven filtering, voice interactions, an AI-enabled assistant that can navigate users through cities giving them spatially aware information. Iteration for prototyping and user testing occurred for usability and efficacy. The project provides evidence that spatial computing supplemented with intuitive UI can help in minimizing cognitive load and can prove to be beneficial in decision-making for renting. It’s also an example of how AR wearables can help you translate digital info into real world autonomy. As different kinds of extended reality (XR) allowing people to access these become more mainstream. NestView serves as a great example of how Immersive Design could be used to tackle a real-world challenge in the real estate space
Evaluating AI-Powered Predictive Analytics for Public Transport Demand in Dubai
No full textThis thesis examines the application of AI-driven predictive analytics to model demand for metro and tram rides in Dubai, utilizing machine learning models in SAS Viya, as Dubai advances its innovative city initiatives. Research focuses on how seasonal variations, the use of metro lines, and types of communities affect trends in passenger flow. Gradient Boosting emerged as the most reliable predictive model, achieving the highest KS values in all scenarios. It effectively captured the increase in ridership during winter on the Red Metro Line and identified summer declines on the Green Line. The findings highlight significant seasonal fluctuations in residential neighborhoods, underscoring the need for adaptable infrastructure planning and scheduling. In contrast, the commercial and industrial zones exhibited more stable patterns of ridership. By incorporating seasonal comparisons, line-specific segmentation, and spatial analysis, the study improves the understanding of how environmental and geographic factors influence the number of riders. Gradient Boosting’s ability to capture these variations provides a foundation for strategic, AI-driven resource allocation and transit scheduling. This research supports Dubai’s smart mobility goals by demonstrating the practical application of AI in improving public transportation systems. By facilitating more accurate predictions and segmentations of ridership demand in various contexts, this study provides actionable information for transit authorities and urban planners seeking to boost efficiency, sustainability, and user satisfaction in one of the fastest-growing cities globally
Teaching and normalizing digital accessibility for students in the chemical sciences
Get PDFDigital accessibility is the inclusive practice of ensuring that all people can equally access digital information. However, this good practice has yet to be fully embraced by practitioners and educators in the chemical sciences. This article will describe and explain the key requirements of digital accessibility and how they apply to educators in the chemical sciences, with the emphasis on supporting visually impaired students; it will illustrate a case study of training students in adopting digital accessibility good practices in order to enable a culture change of greater inclusivity for the next generation; and finally illustrate low cost laboratory adaptations that were found to aid a visually impaired student in the laboratory
Bioinformatics Pipeline for Gene Regulatory Network Discovery: A Case Study with Exercise Response
Get PDFGene regulatory networks are at the core of many biological processes, dictating how a wide range of genes, and their products, interact with one another. Gaining further understanding on how these networks are structured and how they function within the body can aid in deriving meaningful insights on the body’s inner-workings when exposed to any number of conditions. Unfortunately, the information gathered within some studies may be lacking in, for example, sample size or measurements taken at multiple time-points. This work addresses these issues and outlines a pipeline which navigates these smaller collections of data in order to still extract meaningful insights – particularly on gene regulatory networks for a dataset involving various exercise interventions. The work described below demonstrates how a number of statistical techniques, tools, and literature review can yield numerous findings to include: genes relevant to exercise, the relationships between those genes and how they vary across the exercise interventions, and the longevity of these gene regulatory networks
Color Constancy in Virtual Environments with Head-Mounted and Flat-Panel Displays Across Different Illuminants and Lightness Levels
Get PDFVirtual reality (VR) is commonly used as a tool for enhancing immersion within virtual environments, with one of its primary goals being to approximate real-world perception as closely as possible. In everyday life, variations in lighting conditions significantly influence how colors are perceived across different times and settings. However, the human visual system compensates for these changes, maintaining relatively stable color perception; a phenomenon known as color constancy. This process is driven by chromatic adaptation, the visual system’s adjustment to changes in illumination. Despite the growing use of VR, little research has explored how color constancy operates in virtual environments viewed through a head-mounted display (HMD), particularly in comparison to conventional flat-panel displays. To address this gap, we conducted a psychophysical achromatic adjustment experiment to compare color constancy between VR and flat display conditions. We also examined the influence of secondary factors such as illuminant chromaticity and target lightness, testing five illuminants (white, red, green, blue, yellow) and three lightness levels (L* = 40, 55, 70). Our findings reveal that color constancy is significantly higher in VR than on flat displays across all tested conditions. We also observe significant differences in constancy between illuminants, with the green illuminant yielding the highest constancy and red the lowest. Additionally, higher target lightness levels led to lower constancy. By directly comparing VR to flat displays, our study offers evidence that perceptual measures like color constancy may serve as quantifiable indicators of immersion. This has important implications not only for refining theories of color constancy within virtual environments, but also for the development of more immersive virtual experiences
ProtPen: A pipeline that combines sequence- and structure-based approaches to predict protein functions
Get PDFProteins of unknown function represent a significant gap in our understanding of biological processes, with many organisms, especially prokaryotes, harboring large portions of their proteomes that remain uncharacterized. For example, in Pseudomonas aeruginosa, a major human pathogen, a large proportion of proteins are categorized as unknown or hypothetical in function. The precise number varies across species, but studies have shown that up to 30–50% of proteins in bacterial proteomes can lack functional annotations. Addressing this gap is critical to understanding the biology and pathogenicity of such organisms. Recent advancements in computational tools, particularly those in sequence and structure-based prediction, offer new opportunities to annotate these proteins of unknown function. Here, we present a computational pipeline, ProtPen, that integrates eggNOG-mapper for sequence-based functional annotations and Foldseek for structural comparisons using AlphaFold-generated models. ProtPen begins by using FASTA sequences to generate functional annotations from eggNOG-mapper, followed by the retrieval of AlphaFold protein structures from UniProt. These structures are then analyzed using Foldseek to identify structural homologs. By combining these results, the pipeline improves the accuracy and comprehensiveness of functional predictions. We applied this pipeline to quantitative proteomics datasets from P. aeruginosa strains PAO1 and LESB58, where a significant proportion of differentially abundant proteins were of unknown or hypothetical function. In the PAO1 strain, 7 out of 21 proteins were of unknown function, and the pipeline provided annotations for 5 of these. In the LESB58 strain, 28 out of 66 proteins were of unknown function, and the pipeline provided annotations for 20 of these. Our findings demonstrate that combining sequence and structure-based approaches offers complementary insights into protein function. When integrated with quantitative proteomics data, the pipeline provides functional insights into a large fraction of previously uncharacterized proteins, for which their significant proteomics results already demonstrated importance in antibiotic resistance. Notably, this versatile pipeline is easily extendable to new annotation tools and applicable to protein sequences from a wide range of organisms and datasets
Chinese Creative Writing Studies
Get PDFReview of Rebecca Leung Mo-Ling, editor. Chinese Creative Writing Studies. Springer Nature, 2023. 175 pages
Glue Laminated Timber: A New Way of Framing the Residential Architecture of Upstate New York to Improve Thermal Performance & Energy Use Intensity
Get PDFThis thesis addresses the issue present within the current construction techniques for the residential architecture of Upstate New York. Modern platform framing has long been the dominant fashion for constructing homes in the United States, however the nature of this construction is instilled with the issue of thermal bridging, where the wood studs and structure allow heat to escape due to gaps in the continuous insulation. This study proposed a new way of framing residential architecture for the upstate region of New York, relying on glue-laminated timber (GLT) to replace dimensional sawn lumber. GLT has increased structural strength and stability over sawn lumber, decreasing the need for vertical studs. This research focused on assessing whether the GLT prototype assembly can mitigate thermal bridging in residential architecture and pose as a long-term alternative to platform framing techniques. Following a structural analysis, models simulated a typical single-family residence in the city of Glens Falls New York. Models of four conditions were created, platform frame, staggered stud, structural insulated panels (SIPs) and the prototype glulam assembly. These models were simulated using thermal analysis tools to determine overall U-value of the wall and roof assemblies. This data was used to derive the yearly energy use intensity (EUI) and yearly utility cost for each assembly. The glulam prototype was determined to be the most efficient envelope, achieving a 3% reduction in EUI (kBtu/ft2/year) and a 2.3% decrease in yearly utility cost (USD) when compared to platform framing. Overall, it was determined that GLT can improve thermal efficiency in residential building envelopes by decreasing energy consumption and utility costs
Sustainable Retrofits: The balance between energy efficiency and embodied energy
Get PDFThis thesis looks at how to balance energy efficiency with embodied energy in small commercial retrofits, focusing on restaurant buildings. Rather than tearing down older buildings and starting from scratch—something that wipes out all the embodied energy already invested—retrofitting can improve performance while preserving what\u27s already there. Using EnergyPlus for operational energy modeling and the Athena Impact Estimator for embodied energy analysis, this study compares different retrofit scenarios. These include retrofitting one, two, or all three major assemblies (walls, roofs, and floors) to see how each combination affects total emissions over time. The results show that not all retrofits are created equal. Wall retrofits had the biggest positive impact, even outperforming new construction after 10 to 20 years. On the other hand, some upgrades—like floor retrofits—increased cooling demand and overall emissions. The takeaway is that more retrofitting doesn’t always mean better results. The key is knowing which upgrades give you the best return without adding unnecessary embodied carbon
On Studying Transformer Networks for Volume-to-Surface Registration of Inhomogeneous Soft Bodies for Liver Laparoscopy
Get PDFAn important practical consideration in laparoscopic liver surgery is the limited visual information relative to open surgery. In laparoscopic interventions, the surgeon’s view of the liver surface is generally limited to the scene provided by a single scope with a narrow field-of-view. This limits the ability to navigate towards internal lesions previously identified through pre-procedural imaging. Surgical navigation during laparoscopy could be enhanced by registration of full preoperative liver models derived from pre-procedural imaging scans onto the partial laparoscopic view of the liver. This entails both a rigid registration to match the liver surface view to the preoperative volume, as well as a nonrigid registration to correct for the deformation between the pre- and intraoperative scenarios. Prior work has demonstrated the feasibility of data-driven methods for both tasks. However, both registrations are impeded by intraoperative occlusion, limited surface landmarks, and movement of the liver due to abdomen insufflation and patient breathing. In this work, we extend state-of-the-art deep learning frameworks for the task of nonrigid registration. In particular, we investigate the use of vision transformer attention blocks within a U-Net structure to improve the prediction of a displacement field between the preand intraoperative surfaces. We also investigate the robustness of these networks by creating novel training data with imperfect rigid registration and inhomogeneous mechanical properties. Our results indicate that the addition of rigidly deformed training data improves network performance regardless of rigid transformation in the test set. Specifically, we show that a network trained with rigidly transformed data can achieve displacement prediction errors of less than 5 mm on a simulated liver task. Contrary to expectations, the use of transformer architectures and training on inhomogeneous data each reduce network performance across nearly all cases. This work highlights the advantages of a neural network registration method: it requires no knowledge of boundary conditions, has no reliance on manually-tuned parameters, and demonstrates robustness towards sub-optimal prior rigid registration. This work expands the corpus of research backing data-driven volume-to-surface registration as a potentially powerful tool in the advancement of surgical navigation