22505 research outputs found
Sort by
The Transition of the ICT Sector in Kosovo from Service-Based Companies to Product-Based Companies
Full text linkThe capstone project will discuss the barriers and opportunities of the ICT sector of Kosovo to shift its business models into the product-based models, with an emphasis on the economic, social, and cultural determinants of the shift. ICT sector in Kosovo has been dominated long by outsourcing and service provision, yet the world trends and local ambitions require a shift towards innovation-based scalable product models that would create sustainable growth and competitiveness. The study is of qualitative nature and it takes the form of semi-structured interviews with 13 main stakeholders representing government, industry, education and cultural sectors. These interviews provide a thorough understanding of the systemic barriers, including funding constraints, skill gaps, and risk-averse cultural attitudes. Secondary data, including policy documents and international case studies, is used to contextualize these findings within global best practices, particularly from Estonia and Ireland. The results indicate that although Kosovo has excellent technical skills and a young, vibrant workforce, its ICT sector experiences a serious challenge in shifting towards product-based models. Financial constraints, weak industry-academia partnership, and under-valuing of entrepreneurial ventures in society were identified as the key limiting factors. Nevertheless, the opportunities like the development of public-private partnerships, the involvement of the Kosovo diaspora, and corporate governance frameworks have been mentioned as the ways of opening the potential of the sector. This study acknowledges limitations, such as the qualitative focus and potential stakeholder biases. Future studies may also build on these results by conducting quantitative studies or focusing on a particular ICT subsector such as SaaS or artificial intelligence. With the help of the identified barriers and opportunities, Kosovo will be able to make its ICT sector a competitive actor in the global technological market
Future Scenarios of Adoption Sustainable Materials by Construction Firms in the United Arab Emirates
Full text linkThe construction industry accounts for 39% of greenhouse gas emissions, and change is required, especially with the Net Zero initiative. UAE construction companies must evaluate their current positions and update their strategies to support government regulations. Building futuristic communities is crucial for expanding the built environment within the country; however, the types of materials used should account for two aspects: sustainability and climate resilience, and both must be considered whenever these construction companies formulate their strategic response. Hence, scenario planning significantly provides valuable insights to decision-makers in construction companies’ strategic departments to envision four plausible scenarios of the external environment of the construction industry in ten years. This will help them update their strategies to benefit from upcoming opportunities and prepare for the worst-case scenarios. This thesis relied heavily on literature reviews, digital libraries, and market and government announcements to collect valuable data to identify the drivers of change expected to shift the future. At the end of this thesis, four scenarios will be constructed, each requiring a different strategic response in which construction companies could be winners or losers depending on their positions and mitigation plans. Lastly, the drivers of change, uncertainty, and impact levels could change significantly based on new market trends; this must be reviewed and updated regularly, impacting future pathways
Narrow Band Random Vibration Reliability of a Doped Near Eutectic Tin-Bismuth-Based Alloy with a Large Fine-Pitch Ball Grid Array Device
Full text linkThis thesis investigates the vibration reliability and microstructure of two low temperature lead-free solder alloys. The experimental design involved a 2x2 factorial setup, with solder paste type and time above liquidus (TAL) as the primary factors. The solder pastes tested were Indalloy 281 and Indalloy 303, with TALs of two minutes and three minutes. Indalloy 281 is a eutectic Sn42/Bi58 solder and Indalloy 303 is a doped near eutectic Sn-Bi solder. The objective of the study is to identify how reliability changes between solders and the time above liquidous and to also examine the microstructure and the significant factors that impact vibration reliability. Vibration is a common failure mode for the automotive industry, however, there is limited published research on low temperature solder vibration reliability. The reliability of the solder joints is evaluated using Weibull analysis, linear regression is used to identify the significant factors, and cross-sectional failure analysis is used to evaluate the microstructure. Results indicated that Indalloy 303 exhibited superior vibration reliability compared to Indalloy 281 when a shorter TAL was used. The longer TAL generally improved reliability for both solder types, with a more pronounced effect on Indalloy 281. Microstructural analysis revealed that the dopants in Indalloy 303 resulted in Ag3Sn phases in grain structure, contributing to its enhanced performance. The Weibull analysis demonstrated that the failure modes for Indalloy 281 were primarily due to inherent material limitations, whereas Indalloy 303 showed improved resistance to vibration-induced fatigue. Linear regression analysis identified significant factors influencing solder joint reliability, including GRMS mean, board weight, and time since manufacturing. This study observes that Indalloy 303 presents a promising alternative to traditional eutectic solders, offering improved reliability and potential for broader application in the electronics industry. There are some concerns with the one test condition whose data did not appear like the other data which indicates there is possibly a manufacturing or vibration testing concern for that data. From this study there are many future research opportunities including analysis of this experiment that has not been done and running a large sample size experiment. Studies should explore other low-temperature solder alloys and their performance under various environmental conditions to further advance lead-free soldering technologies
AN ANALYSIS OF DC-DC CONVERTER EFFICIENCIES
Full text linkWith the increase in technology used in everyday life, the need for DC-DC converters has continued to increase. The efficiency of these converters is an essential parameter as an increase in a few percent can increase battery life by hours. There are several different factors that effect the efficiency of converters ranging from temperature, component selection, quiescent current and more. In this paper, a series of eleven DC-DC converters will be designed and tested in order to observe how the efficiencies and other parameters compare to the values provided in this datasheet. For this investigation two different PCBs were designed, one containing converters designed using the assistance of TI WEBENCH to select the semiconductor chip and necessary passive components, and one with a series of evaluation board modules. An additional three fully assembled evaluation boards were tested for a total of five buck converters, five LDOs and one boost converter. Through this investigation significant differences were found between the expected values provided in the datasheet and the measured values. These discrepancies can be accounted to a wide variety of sources such as component selection, PCB layout and precision of the measurement. This effort was documented in detail for the design, testing and analysis of the results
Machine Learning for Complex Networks: Generalization, Fairness, and Model Selection in Graph-Based Systems Across Social and Biological Domains
Full text linkNetworks serve as a fundamental framework for modeling complex systems across diverse domains, from biological interactions and social structures to economic markets and information networks. Machine learning has significantly enhanced the ability to extract meaningful patterns from graph-based systems, improving predictive accuracy in applications such as market design, recommendation systems, and biomedical discovery. Despite the successes of machine learning in analyzing complex networks, several fundamental challenges remain. A major issue is the reliance on simplistic models for theoretical analysis, simulation, and benchmarking, which often fail to capture the complexities of real-world networked systems. Many theoretical studies focus on oversimplified generative models that assume uniform randomness in edge formation, preference rankings, or network evolution, limiting their applicability to real-world problems. While these models provide analytical tractability, they often overlook the structured dependencies and correlations that shape real-world interactions. Similarly, the simulation and benchmarking of machine learning models for networks frequently rely on synthetic datasets that do not reflect real-world network structures. Many standard benchmarks use random graph generation or random edge masking for evaluation, disregarding the fact that real-world networks evolve dynamically and exhibit structured missingness. For example, in link prediction tasks on biomedical graphs, edges are often removed at random for training and evaluation, despite the fact that in practice, new edges are discovered sequentially over time, requiring models to generalize to future observations rather than random missing links. Similarly, in stable matching problems, synthetic preferences are often generated using uniform distributions, ignoring the empirical correlation patterns observed in real-world matching markets, such as labor markets or school admissions. Addressing these challenges requires a shift toward realistic domain-specific data modeling, evaluation methodologies that reflect temporal and structured missingness, and models that are designed for deployment in dynamic, real-world networks. This dissertation addresses these challenges by: (1) Investigating fairness in stable matching with correlated preferences, revealing how preference asymmetries shape fairness outcomes and proposing efficient stability-preserving fairness-aware solutions. (2) Developing a principled model selection framework for signed networks, introducing GRASMOS, a maximum likelihood-based approach to infer realistic sign assignment patterns in gene regulatory networks. (3) Improving evaluation methodologies for link prediction methods by incorporating temporal graph evolution, reducing generalization gaps in biomedical link prediction. By advancing fairness-aware algorithms, realistic generative models, and improved evaluation methodologies, this thesis contributes to more reliable, interpretable, and generalizable machine learning methods for structured decision-making in graph-based systems across social and biological domains
Reducing CO2 Levels in a Classroom Through the Use of Biofilters
Full text linkWe spend 80% of our time indoors and are seeing an increase in indoor pollutants (Sharma et al., 2022). With the push to create more energy efficient buildings, the envelopes are becoming more airtight. As a result, indoor air quality (IAQ) has decreased due to the inability of indoor pollutants to escape from the building as easily through cracks and seams. (Irga et al., 2018; Satish et al., 2012). To date, there is a lot of existing research in regard to the role biofilters play in improving IAQ. Research ranges from having examined specific parts of the plants, soil conditions, and the display of the plant, such as potted or vertical on a green wall, to see how those factors contribute to the plants’ ability to filter air. Majority of these experiments regarding improving indoor air quality (IAQ) through the use of biofilters have taken place in a controlled laboratory setting, leading to a lack of infield testing. Therefore, this paper addresses the research gap of infield biofilter testing. The goal of the four-week experiment is determine if the increased use of biofilters such as Boston ferns, spider plants, and Jade plants within the classroom will help reduce the CO2 levels of the classroom. Results indicate that while there was no significant change in CO2 levels with the placement of biofilters in the classroom, there was a significant impact on occupants’ psyche such as attentiveness, energy levels, mood impact, and perceived changes in IAQ
Using Generative AI for Tutoring Data Science
Full text linkA large increase in the use of Generative AI has been observed in the last few years. Data science is also a rapidly growing field with a high demand for skilled professionals. The goal of this thesis is to explore the potential of Generative AI, specifically ChatGPT, in facilitating data science education. The focus is on how ChatGPT can be used as a tutor to help solve practical exercises. The capabilities of Generative AI were explored along with the comparison of a few different models in terms of data science. Exploratory analysis was conducted to compare Generative AI models and choose the better one for the thesis. The best practices for solving practical exercises using Generative AI were also explored. The prompt engineering practices for solving practical exercises have also been explored and described in this thesis. The effectiveness of using ChatGPT as a tutor of data science has also been evaluated in three different ways. First, a series of sessions were created with ChatGPT to help solve and explain data science concepts in a structured way and the accuracy of these answers was analyzed. Second, a study was conducted to see how helpful ChatGPT is in helping participants solve data science questions. The third approach was using another Generative AI model, Claude, to test how ChatGPT acted as a tutor for undergraduate students. It was found that ChatGPT provides more factually correct answers as compared to Gemini and is better at solving problems and explaining concepts as compared to GitHub Copilot. There are limitations to ChatGPT when it comes to computations and solution building for data science, but its use can facilitate the learning process of students. Topics like schema building, data creating, query writing, normalization, itemset mining, and clustering can be learned and understood with ChatGPT. Using it for educational purposes will facilitate faster and better learning for students as compared to scenarios where its use is prohibited. ChatGPT can be used for solving questions and explaining answers. Students can work on questions step by step with the help of ChatGPT and learn the process of solving the questions. Some questions can be solved easily with simple prompts while some require more structured prompts. The results of our evaluations are mostly positive with a few limitations. The results show that ChatGPT can be used as a tutor for learning data science but can not be the only source of learning. Some guidance or knowledge is needed for better use of the Generative AI. Our main takeaway is that Generative AI can not substitute teachers but can act as a personalized tutor for each student. It can explain the solutions given in the textbooks in more detail and can also help with error resolution. A large number of participants also said that they are likely to use ChatGPT as a data science tutor in the future. In conclusion, ChatGPT has the potential to revolutionize data science education by acting as a personalized tutor, enhancing the learning experience, and bridging gaps in understanding complex concepts
Classification of Linear Systems of Equations for Quantum Computing Implementation
Full text linkDrawing nearer to an error-corrected era of quantum computing, it is necessary to understand the suitability of certain post-NISQ algorithms for practical problems. One of the most promising, applicable, and yet difficult to implement in practical terms is the Harrow, Hassidim and Lloyd (HHL) algorithm for linear systems of equations. An enormous number of problems can be expressed as linear systems of equations, from machine learning to fluid dynamics to electrical circuit analysis. However, in most cases, HHL will not be able to provide a practical, reasonable solution to these problems. This work seeks to determine whether problems can be labeled as suitable or unsuitable for HHL implementation using machine learning classifiers when some numerical information about the problem is known beforehand. It is demonstrated that training on a data distribution that is sufficiently representative of the target problem space is critical to achieve good classifications of the problems based on the numerical properties of a system’s coefficient matrix. Accurate classification is possible using multi-layer perceptrons, though only with careful design and selection of the training data distribution and classifier optimizations
A Remote Sensing-Based Evaluation of Road and Electricity Infrastructure Expansion Effects on Development in Sub-Saharan Africa
Full text linkSub-Saharan African countries consistently rank low on human development indicators despite decades of international aid and domestic investment. To address these challenges, governments across the region are allocating substantial portions of their national budgets to infrastructure expansion, viewing it as a catalyst for accelerated socio-economic development. However, given severely constrained budgets, these nations often face trade-offs between infrastructure development and investment in other important sectors such as healthcare and education. This highlights the need for infrastructure investments to deliver their anticipated benefits. This dissertation examines the relationship between infrastructure development and socio-economic development in Sub-Saharan Africa, with a particular focus on electricity and paved road access as enablers of development. The research specifically investigates how the expansion of these infrastructure influence two important aspects of development: (1) the adoption of irrigation in agricultural systems and (2) patterns of land use and land cover change. A key methodological contribution of this work is its approach to addressing the region’s data accessibility challenges. Previous research examining infrastructure’s role in development has been constrained by this data scarcity, forcing studies to either focus narrowly on individual villages, communities, or specific projects, or to rely on aggregated national-level statistics that conceal local variations. This dissertation bridges this gap by leveraging remote sensing and machine learning techniques to develop medium- to high-resolution, multitemporal maps of infrastructure access. This multi-scale approach enables assessments across household, community, district, and national scales. The resulting datasets enable the tracking of spatiotemporal changes in electricity and all-weather road infrastructure coverage, providing an understanding of when a particular community got access to the infrastructure. Beyond its research applications in this dissertation, this methodological framework offers a cost-effective, replicable approach for monitoring infrastructure development in data-scarce environments that can be adapted by researchers and policymakers. The findings reveal that electricity and paved road access can be mapped reliably with freely available nighttime light and surface reflectance satellite images, achieving balanced accuracies as high as 86% and 95% respectively. We found a weak correlation between infrastructure access and irrigation adoption in farmlands in northern Ethiopia. Instead, irrigation efficiency measures and crop types emerged as stronger determinants of irrigation adoption. Lastly, we found that access to electricity significantly increased cropland at the expense of rangeland and forest in Rwanda. The dissertation is structured into five main interconnected chapters, each addressing specific research questions while contributing to a deeper understanding of the relationship between infrastructure and development. The results offer insights for policymakers seeking to maximize returns on infrastructure investments, particularly by developing approaches for targeting interventions to improve agricultural processes
Waylight: Mapping the Journey of Faith
Full text linkWaylight is a digital learning platform that reimagines how we explore the life of Jesus through maps, movement, and meaningful interaction. Designed to bridge the gap between static Bible study tools and the physical settings of Scripture, the platform allows users to follow curated journeys such as Miracles or Timelines across ancient Israel. The project began with experimental designs using three-dimensional terrain and literal religious symbols. These early versions lacked cohesion and felt visually heavy. As the work progressed, the design shifted toward a more refined two-dimensional aesthetic inspired by textures like layered paper and the craft of carpentry. This direction maintained a sense of warmth and symbolism while supporting a clean, modern interface. The final map was built in Illustrator and prototyped in Figma, focusing on clarity, structure, and spatial storytelling. Each location on the map includes a modular content card with selected Scripture, contextual background, and optional reflections. The interface is designed to be accessible and calm, using a neutral color palette, clear typography, and thoughtful pacing to support users across a wide range of ages and learning styles. Public feedback at Imagine RIT confirmed the system’s educational value and cross-generational appeal. Visitors expressed interest in using it for personal study, youth groups, and classrooms. The long-term vision for Waylight is to expand beyond the life of Jesus to include both Old and New Testament journeys. Future development could add devotionals, quizzes, and study guides. While artificial intelligence supported early research and prototyping, final content would be created by pastors, historians, theologians, and educators to ensure accuracy and depth. Waylight demonstrates how visual communication and interaction design can support meaningful engagement with Scripture through a clear and spatially organized experience