Carleton University Institutional Repository
Not a member yet
20816 research outputs found
Sort by
Exploring the Influence of Avatar Design on Patient Self-Disclosure within Virtual Therapeutic Environments
No full textThis thesis explores the impact of avatar design on patient self-disclosure within VTEs. My research explores how the different types of avatar anthropomorphism can influence the frequency and type of self-disclosure during therapy. I used a mixed methods approach that included quantitative, subjective quantitative, and qualitative data. Participants interacted with six different custom virtual avatars in a virtual environment. Their self-disclosure were recorded and analyzed. This was followed by a survey aimed to better understand the user’s perspectives on virtual avatars. The survey helped understand the scores from the study and further improve its validity. My research explores the broader implications of avatar design in virtual environments, focusing on spatial arrangements, color, and lighting. The results provide insights at optimizing avatar and environment to enhance therapy. This thesis contributes to the growing field of VR therapy by understanding virtual avatar design in meaningful therapeutic interactions and helping improve mental health
Streamlining Test Processes with Applied Artificial Intelligence: Integrating and Analyzing Test Data for Cost Reduction in 5G and Beyond Software - an Industry Perspective
Get PDFIn the rapidly evolving telecommunications industry, organizations face the challenge of ensuring that a diverse range of products perform effectively across various configurations. The emergence of new technologies, particularly in radio access and frequency bands, complicates testing and validation processes. Companies must strike a balance between comprehensive testing to guarantee quality and the constraints of time and resources, necessitating innovative approaches to enhance testing strategies. This thesis proposes a holistic framework that leverages insights from historical test data and domain expert knowledge to derive actionable insights, streamline processes, improve compatibility, and reduce costs while maintaining product quality in a competitive market. We employ various reverse engineering techniques, including deep learning-based natural language processing, to navigate the complexities of testing artifacts and extract a machine-readable mapping between information about test cases and the functions of the system under examination. This machine-readable feature mapping is then fed into an interpretable machine-learning engine to extract actionable insights. The research is structured into two phases: Phase 1: Data collection procedures focus on extracting information from diverse data sources using API access, analyzing test complexity, and mapping features to test cases through deep learning techniques. It involves identifying data sources and creating a mapping between features and test cases. Phase 2: Data analysis procedures analyze the feature mapping to extract actionable insights using interpretable machine learning techniques. This phase classifies mappings based on a classification target, enabling predictions, such as which product configuration combinations new test cases should prioritize. The combined insights from both phases contribute to developing an AI/ML-driven streamlining procedure that enhances testing efficiency through: -Optimized Configuration Selection: Classifying test cases by execution time helps identify configuration combinations that yield the shortest execution time while considering contextual relevance and interdependencies. - Effective Test Effort Allocation: Analyzing classified test cases with a specific classification target identifies test frequency imbalances and informs better resource allocation. -Test Gap Analysis: Data insights reveal gaps in testing coverage, highlighting under-tested configuration combinations. Through this structured approach, this thesis aims to bridge knowledge gaps, mitigate uninformed decision-making, and address challenges stemming from data silos and heterogeneity in test information
Domain Specific Commodity Classification in the Domain of Central Bank Economics
Get PDFIn the field of economics, analyzing market news for commodities like oil is crucial for forecasting trends and making informed decisions. The sheer volume of news data requires efficient methods for sentiment analysis. This thesis explores the use of language models for sentiment analysis within the oil commodity market, focusing on extracting information related to price, supply, and demand dynamics from daily news. The study investigates the efficacy of zero-shot and few-shot learning, along with the use of adapters for contin- uous training, in both small and large language models. It is hypothesized that few-shot prompt engineering offers a cost-effective and efficient solution for sentiment analysis in this context. The research examines the performance of various models, including those trained on domain-specific datasets and those continuously trained with adapters. The findings contribute to developing more accurate and efficient tools for economic analysis and forecasting, while also considering the environmental impact of different techniques
Rekindling Lost Communities: Reimagining How to Rebuild Northern Communities After Wildfire Disasters
Get PDFFew impacts from Climate Change are as devastating to communities as wildfires. As recently seen in Jasper, Alberta and Los Angeles, California, wildfires can devastate both remote communities and densely populated urban areas. Drawing on recent case studies in Canada and the U.S., the thesis identifies the challenges communities face when responding to wildfire-related disasters. This thesis uses Yellowknife, NT, as a hypothetical case study given its near-catastrophic wildfire experience in 2023. This thesis examines approaches to reconstruction, focusing on rapidly deployable temporary housing that can transition toward permanent use. Among the topics covered are proactive measures that can enhance the resilience of communities like Yellowknife when facing such disasters. These include the use of modular and prefabricated components for the construction of replacement housing
Unsupervised 3D Shape Parsing with Primitive Correspondence
Get PDF3D shape parsing, the process of analyzing and decomposing a 3D shape into components, is an important task in computer vision and graphics. Shape parsing approaches include segmentation and approximation methods. Approximation methods often represent shapes with a set of primitives fit to the shapes, such as cuboids, cylinders, or superquadrics. However, existing approximation methods typically rely on many initial primitives and aim to maximize coverage of the target shape without accounting for correspondences among primitives. This thesis introduces a novel 3D shape approximation method integrating reconstruction and correspondence into a single objective, ensuring consistent approximations across shapes. Experimental results demonstrate that incorporating correspondences improves consistency and approximation quality when using fewer primitives. Moreover, although correspondences are estimated in an unsupervised manner, our method effectively leverages them to reduce reconstruction error
Viviendo en la Frontera: Exploring the Racialized Experiences of Queer Latinas in Ottawa
No full textIn this dissertation, I explore how queer Latinas living in Ottawa navigate, negotiate, and challenge the dominant power structures of whiteness in the city, including its queer white environments. Based on eighteen months of ethnographic fieldwork in the queer bar, The Hot Room Bar and Nightclub in the ByWard Market, my work focuses on Latinas’ queer practices, including as they relate to gender performance, migration, identities, work, education, health care, family, friendships, sexual relationships, marriage, motherhood, language and mass media. At the same time, I refer to Latinas’ life experiences in Ottawa and the ways in which they engage with the ideology of multiculturalism. This thesis takes up Anzaldúa’s theory of the borderlands to reveal the contradictory ways in which queer Latinas navigate, negotiate, engage, and challenge their social border crossings. Through borderlands, the dissertation focuses on the shifting processes experienced by Latinas post-migration. As well, it refers to Latinas who despite not having experienced migration, still experience racialization because of their Latin American or Caribbean backgrounds. I illustrate how these women engage and use social mixing to reinvent themselves with the purpose of living meaningful lives post-migration. It is within this reinvention that these women are also able to break up the binary thinking of being Latin Americans or Canadians and instead, create new identities for themselves. I examine the complex ways in which queer (migrant) Latinas experience different forms of racialization in the city and how this racialization becomes part of the shifting process of Latinas’ self-reinvention post-migration. I track stereotypes as one of the forms in which Latinas experience Ottawa and while some Latinas experience them in harmful ways, other Latinas use stereotypes in creative ways for social and economic mobility, and some even to succeed on television
Acoustic Approaches to Locating Bat Roosts and Identifying Drivers of Bat Activity in Canada’s Largest Conurbation
Get PDFAcoustic survey methods are used to track bat populations declines and to address ecological knowledge gaps. I tested two applications of acoustic monitoring in Toronto’s Rouge National Urban Park. First, to identify whether acoustic data can locate bat maternity roosts, I tested whether Eptesicus fuscus activity detected by acoustic monitors is predicted by a monitor’s distance to known roosts, and time elapsed since sunset. Time and distance predicted bat activity, but effect sizes were small, limiting the precise location of roosts based on acoustic data. Second, I tested how the amount of available bat habitat affects bat activity, using 7 years of acoustic data. Most landscape features I tested did not predict bat activity, although activity of northern myotis (Myotis septentrionalis) was positively associated with forest cover. Temperatures above 10°C increased bat activity, while precipitation reduced activity. This thesis highlights both the strengths and limitations of bat acoustic surveys
Virtual Prototwinning: Exploring Digital Twin Development in Virtual Reality Using Numena’s Space Elevator and a Raspberry Pi to Design and Build a Birdhouse Digital Twin
No full textThis thesis explores Virtual Prototwinning (VP)—a workflow that integrates designing in Virtual Reality (VR) to build Digital Twins (DTws). While DTws are widely used in industries like manufacturing and urban planning, they are often an afterthought in architecture and design. VP uses VR as a design tool for DTws. A case study on birdhouse virtual design demonstrates how individuals can engage with DTw creation. Findings from the conducted user study suggest that VP enhances user engagement, fosters iterative decision-making, and improves approachability of DTw technology. This research highlights the potential for democratizing the development and use of DTws, shifting from top-down implementations to participatory, community-driven approaches. By demonstrating VP’s ability to bridge digital and physical spaces, this thesis lays the foundation for broader applications. As these systems grow in scale and complexity, thoughtful governance will be crucial to address issues of data ownership, accessibility, and long-term accountability
Mourning as a Spatio-Temporal Practice: Towards a Theory of Preservation during a Time of Destruction in Bilād al-Shām
Get PDFIn modern history, western colonial and imperial necropolitical violence has perpetuated a state of mourning on the lands and people of Bilād al-Shām. This thesis defines this state as necropolitical mourning—recognising and condemning its causes and perpetrators—and explores the question of mourning as both an imposed state and—in its reclamation—as one of transformative potentiality. By adopting preservation as methodology, this thesis frames the practice of mourning as one grounded in remembrance and unrestricted in imagination; a refusal of oppression and an insistence upon personhood. If life and death can be defined in both the physical and metaphysical realms, then mourning exists across both; its transformative potential—preservative and revolutionary—embedded in its shaping of conceptions of home-land, memory, and time. This thesis confronts these seemingly antithetical terms, offering an understanding of mourning, preservation, and resistance, not as distinct terms, but as of one another