19468 research outputs found
Sort by
Roads and Corresponding Travel Time to Markets: Assessing Climate Vulnerability in Nepal
No full textRoads exist as a physical and theoretical connection between people and places around the globe. In addition to providing a route from one point to another, roads are also an indicator of access to markets and of poverty. However, current road datasets, particularly the Global Roads Open Access Data Set, are out of date or incomplete, necessitating new sources of data for analyses involving road networks. This study explores the relationship between climate change and access to markets in Nepal. We seek to identify isolated communities that are likely to experience detrimental outcomes associated with environmental threats, such as increasing temperatures and unpredictable precipitation. To implement this analysis, we first construct a novel pipeline to retrieve and analyze road data from Open Street Map (OSM). The output of this pipeline is a gridded product that includes information, for each grid cell, on the total travel distance and time to the closest market. By comparing this road data with future environmental change projections, we identify communities that are both geographically isolated and at risk due to impending climatic shifts. This study reveals the high vulnerability of the northwestern region of Nepal, specifically the Karnali province, due to the lack of road access and the extreme climate impacts Nepal is likely to experience in the coming decades. Our study suggests the benefits of applying this vulnerability analysis at a global scale.Data ScienceBachelors of Science (BS
Ecological Niche Modeling, Index Standardization, And Recovery Times For Data-Limited Coastal Sharks
No full textCoastal sharks, as part of the class Chondrichthyes, are one of the most threatened species groups in the world. Observation, assessment, and management are particularly challenging for these species due to their large distributions, complex migratory behaviors, low economic value, slow-growth rates and low-reproducive output (i.e., K-selected) life history strategies, and data limitations. Coastal sharks are also still recovering from overexploitation in the 1970s through 1990s, though recovery times are unknown for many species as most are too data-limited for species-specific stock assessments. Further, anthropogenic climate change is anticipated to impact coastal sharks in various ways and could affect shark management and assessments. Multiple risk assessments have reported that coastal shark species have a high potential to shift distributions in response to climate change. In this study, ecological niche models quantified baseline habitat associations for several small and large coastal shark species along the southeast US Atlantic and Gulf of Mexico across multiple sizes. Output from models indicated that impacts of anthropogenic climate change on environmental conditions, such as increased water temperature, impacted the quantity and quality of available habitat for coastal sharks. Evidence suggested that multiple species could shift distributions north along the Atlantic and/or to deeper offshore waters in the Atlantic and Gulf of Mexico. Additionally, species have the potential to redistribute in response to multidecadal climate variability for multiple species. Changes in species distributions and migration times for coastal shark species likely affect survey catchability and index interpretation. A spatiotemporal index standardization method was investigated for six different data-limited species and compared to two index driven standardization methods currently implemented in stock assessments. Previous studies noting a preliminary recovery of coastal shark populations may have been over-optimistic as only two of the six species examined displayed increasing abundance trends over time. Index standardization methods largely agreed with each other, but positive trends in density and increased variability in density anomalies in the spatiotemporal models suggested a northward expansion or a timing discrepancy between migration onset and sampling efforts for multiple species. Finally, using a stochastic Leslie matrix and Schaefer surplus production simulation, a framework was designed to estimate potential recovery times for data-limited and unassessed species, such as coastal sharks. Twenty-six shark stocks, consisting of small and large coastal sharks as well as dogfishes, were investigated as a case study. Estimated recovery times were compared to a data-moderate age-structured simulation that accounts for nuances specific to shark life history. Recovery times estimated from the Schaefer surplus production simulations were considered viable as a "best-case scenario" since this method routinely underestimated data-moderate simulations. Levels of stock depletion significantly impacted recovery estimates, highlighting the importance of early detection and quick response. Large coastal sharks recovered slower than small coastal sharks and dogfish recovery times varied. While most coastal shark stocks are unlikely to recover within 10 years, conservative estimates for the recovery of collapsed stocks of most large coastal sharks begin around 30 years, coinciding with the duration that shark management has been in place. Collectively, the results of this dissertation help address how anthropogenic climate change impacts coastal shark populations and their management while also providing valuable information regarding the recovery of data-limited and overexploited stocks.Virginia Institute of Marine ScienceDoctor of Philosophy (Ph.D.
"Agent Orange is our Nemesis": The Blue Water Navy Veterans' Battle for Dioxin Compensation Amidst the Ongoing Vietnam War
No full textThis thesis examines the use of Agent Orange in the Vietnam War and its effects on American Blue Water Navy veterans amidst their battle for presumptive service connection to Agent Orange exposure. The first chapter studies the history of the American decision to use Agent Orange, placing this history in the broader context of the Vietnam War. It argues that as long as Agent Orange and its health consequences persist among its victims without proper compensation, the Vietnam War is an ongoing conflict. The second chapter is a focused study on the Blue Water Navy veterans’ battle for a presumptive service connection to Agent Orange exposure, examining the congressional, legislative, and legal debates that surrounded this battle. The third chapter presents three original oral histories of US veterans to chronicle the struggles of the victims while legislative discussions were under way. This thesis introduces a political and social history of US veteran Agent Orange compensation to a Vietnam War literature that largely revolves around military and diplomatic history. It also highlights new oral histories of Blue Water Navy veterans within the literature that does focus on veteran compensation. This thesis uses these oral histories, governmental correspondence, congressional debates, speeches, and more to chronicle US veterans’ battles for health care and compensation for their exposure to Agent Orange in the Vietnam War.HistoryBachelors of Arts (BA
Code Syntax Understanding in Large Language Models
No full textCode can be found at: https://github.com/WM-SEMERU/syntax-error-coleIn recent years, tasks for automated software engineering have been achieved using Large Language Models trained on source code, such as Seq2Seq, LSTM, GPT, T5, BART and BERT. The inherent textual nature of source code allows it to be represented as a sequence of sub-words (or tokens), drawing parallels to prior work in NLP. Although these models have shown promising results according to established metrics (e.g., BLEU, CODEBLEU), there remains a deeper question about the extent of syntax knowledge they truly grasp when trained and fine-tuned for specific tasks. To address this question, this thesis introduces a taxonomy of syntax errors, and a labeled set of LLM generated code containing syntax errors. The taxonomy was organized into Simple and Complex errors, describing the level of structural degradation caused by the syntax errors. We explored these over three different NLP datasets: Mostly Basic Python Problems (MBPP), the Code/Natural Language Challenge (CoNaLa), and HumanEval. With CoNaLa and MBPP having the task of code generation from natural language, and HumanEval having the task of code completion. We ran a total of 4,941 prompts into the Mistral-7b-instruct-v.2 model, and encountered 130 syntax errors, or a 2.6\% error rate. When we restrict the samples to python code only, the error rate increases to 2.9\%. The most common simple error was an extra token, a space, added to the result. The most common complex error broke the assign relationship.Computer ScienceBachelors of Arts (BA
Improving Teachers' Assessment Literacy: The Effects Of A Professional Learning Program
Get PDFTeachers’ assessment practices greatly influence student learning. However, the level of assessment literacy among teachers is inadequate relative to classroom assessment standards and expectations. Assessment literacy includes interpreting results of various assessments, creating assessments that are aligned to learning targets, using assessment results to understand students’ gaps in learning, and adjusting instruction accordingly. This formative program evaluation used mixed methods to examine a professional learning program for teacher leaders in a public school district. Using Kirkpatrick’s model, participants’ reactions, knowledge and skill development, and changes in practice were evaluated to determine the effectiveness of professional learning and inform future professional learning for the district. Findings indicate that teachers enjoyed the professional learning and felt it was valuable to their practice. Participants proficiently identified the process of designing a robust classroom assessment system including developing, using, and analyzing classroom assessment and showed developing skills in identifying kinds of learning targets and defining reliability and validity. Teachers’ approaches to classroom assessment did not change significantly, remaining teacher-centric and endorsing an Assessment for Learning approach before and after professional learning. Appreciative inquiry interviews with teachers revealed changes to classroom practice, confidence in using assessment to improve instruction, and the value of professional learning communities. Teachers conveyed the value of formative assessment in meeting the needs of all students and lowering students’ affective filters. Results support the importance of the socio-cultural context in improving teachers’ assessment literacy and provide a model for effective professional learning that improves classroom practices.EducationDoctor of Education (Ed.D.
A Program Evaluation For The Leadership Academy: A School-Based Program For 12Th Grade Students Who Are At Risk Of Not Graduating High School
Get PDFThe Great Mountain High School (GMHS) started a program to help support students at risk for not graduating high school. The focus of this study was to provide a formative program evaluation of the created program that (a) investigated the fidelity of implementation of the activities and processes of the program, (b) gathered an understanding of the success of the program’s impact on graduation rates, and (c) provided an understanding of the strengths and areas of growth the Leadership Academy. A mixed method, CIPP model, with a pragmatic lens, was used during an analysis of a historic document review, teacher interviews, and student participant surveys. This study found that (a) the academic components of the program were being implemented with fidelity and the community and career components were partially implemented with fidelity when compared to the program’s design, (b) there was no statistical difference between student participants of the program and similar student non-participants, and (c) the success of the Leadership Academy occurred in the value added to the student and student’s perceptions of their life and life after high school graduation. Evidence suggested that the program should continue to be implemented not because of its graduation success but because of the value added to student participants. Further recommendations from this study called for a greater implementation of community-based and leadership-based lessons to increase the fidelity of implementation and more planning time for teachers of student participants to collaborate on more cohesive initiatives.EducationDoctor of Education (Ed.D.
Social Media And Professional Learning
No full textAlthough research indicates professional learning could positively influence both teacher retention and student achievement, many educators continue to express dissatisfaction with professional learning experiences. In a Virginia Department of Education (VDOE, 2019b) survey, teachers at Roth High School expressed issues with misalignment between the content of professional learning offerings and their specific learning needs, a lack of follow-up implementation support, and a failure to see an effect on student achievement. Using Guskey’s (2000) Five Levels of Professional Development Framework, which uses a leveled approached to assessing professional learning, this action research study sought to achieve a better understanding of the potential benefits and potential challenges of using social media for professional learning; how teachers selected professional learning experiences while using the medium; and what additional supports might be needed to support its use. Data collected from a professional learning tracking document, survey, and focus group provided evidence that teachers were pleased with their experiences using social media for professional learning, citing its ability to provide specific learning aligned with their content and usability as two of its most compelling benefits. Because teachers did report some issues with collaboration and information overload, leaders may want to consider how to leverage elements of the 70-20-10 learning model to provide additional opportunities for collaboration, guidance on how to select meaningful learning experiences, and offer follow-up support. Despite these barriers, teachers believe using social media for professional learning to be a worthwhile endeavor, and school leaders should continue to explore ways to leverage it.EducationDoctor of Education (Ed.D.
Milkweeds And Microbiomes: The Role Of Plant Hybridization In Plant Associated Microbiome Community Structure
No full textThis study examines the influence of plant hybridization, environmental factors, and long-term neighborhood effects on A. exaltata and A. syriaca soil and phyllosphere microbiomes. Our approach is an in situ study of milkweed plants in natural hybrid zones and their associated microbiomes. Our approach captured variation in genotype, leaf and soil nutrients, and geographic location to distinguish the varying influences on plant-associated microbiomes. Our analyses demonstrated that inter-species hybridization influences the number of taxa in phyllosphere microbiomes while having a limited influence on the composition of soil and phyllosphere microbiomes. Instead of genotype explaining variation in microbiome composition we found that environmental factors and geographic distance, as a proxy for long-term neighborhood effects, strongly influenced the composition of plant-associated microbiomes. Specifically, soil organic matter, cation exchange capacity, and soil pH were important factors influencing plant-associated microbiomes in milkweed. Overall, we found that plant-associated microbiomes of milkweed plants A. syriaca and A. exaltata are dominated by environmental factors and neighborhood effects, while plant host species and hybridization appears to influence the prevalence of rare or specialized microbes. Future research is needed to understand the specific mechanisms influencing rare microbial taxa in plant-associated microbiomes, as well as the mechanisms of environment factors influencing plant-associated microbiome composition.BiologyMaster of Science (M.Sc.
Emergent Capabilities Of Llms For Software Engineering
No full textA growing interest for Large Language Models (LLMs) is how increasing their size might result in changes to their behavior not predictable from relatively smaller-scaled models. Analyzing these emergent capabilities is therefore crucial to understanding and developing LLMs. Yet, whether LLMs exhibit emergence, or possess emergent capabilities, is a contested question. Furthermore, most research into LLM emergence has focused on natural language processing tasks and models suited for them. We focus on investigating emergence in the context of software engineering, and recontextualize the discussion of emergence in the context of prior research. We propose a multifaceted pipeline for evaluating and reasoning about emergent capabilities of LLMs in any context and instantiate this pipeline to analyze the emergent capabilities of the CodeGen1-multi model across four scales ranging from 350M parameters to 16.1B parameters. We examine the model's performance on the software engineering tasks of automatic bug fixing, code translation, and commit message generation. We find no evidence of emergent growth at this scale on these tasks and consequently discuss the future investigation of emergent capabilities.Computer ScienceMaster of Science (M.Sc.