SHAREOK Repository
Not a member yet
49261 research outputs found
Sort by
Understanding Inter-Group Cooperation Among Far-Right Extremist Groups: Through the Two-Pyramid Model of Radicalization and Textual Analysis
Get PDFAfter the event of the Unite the Right Rally in Charlottesville, NC, it has become clear that extremist white nationalist groups communicate and interact with each other. What is unclear is the criteria they use when choosing to cooperate. I seek to address this gap, by adapting the McCauley and Moskalenko Two-Pyramid Model of Radicalization to the organizations themselves and analyzing whether or not it is opinions or actions that shape one group's opinion of another. To answer this, I apply textual analysis to leaked communications from several white nationalist servers, I isolate statements referencing other white nationalist organizations, code whether or not they are referring to the group's actions or their opinions, and then code whether the statement is negative or positive. I then combined the results of the latter two textual analyses into four different values referred to as statements. With this analysis, I will test the following: H1 - Extremist groups speak more frequently about others' extreme actions than their extreme opinions. H2 - When speaking of other groups' opinions there will be more variation in positive and negative opinion statements. H3 - Groups with a higher number of negative opinion statements than positive opinion statements will have fewer instances of positive action statements compared to negative action statements My analysis shows complete support for H1 and H2 and partial support for H3
Naturalness and Novel Statistical Methods on the String Landscape
Get PDFIn this work, we formalize the concept of naturalness in supersymmetric effective field theories, as well as introduce novel methods for performing statistical analyses in the string landscape. We revisit the various measures of practical naturalness for models of weak-scale supersymmetry (SUSY) including: 1. electroweak (EW) naturalness; 2. naturalness via sensitivity to high-scale (HS) parameters [Ellis-Enquist-Nanopoulos-Zwirner/Barbieri-Giudice (EENZ/BG)]; 3. sensitivity of Higgs soft terms due to high-scale radiative corrections; and 4. stringy naturalness (SN) from the landscape. We debut a new numerical routine for calculating these measures from any SUSY Les Houches Accord file. A vast array of (metastable) vacuum solutions arise from string compactifications, each leading to different 4-d laws of physics. The space of these solutions, known as the string landscape, allows for an environmental solution to the cosmological constant problem. We argue that the landscape favors natural softly broken supersymmetric (SSB) models over particle physics models containing quadratic divergences, such as the Standard Model or unnatural SSB models by presenting a computable measure. An anthropic selection of the weak scale to within a factor of a few of our measured value — in order to produce complex nuclei as we know them (atomic principle) — provides statistical predictions for Higgs and sparticle masses in accord with LHC measurements. The predicted Higgs and superparticle spectra might be testable at HL-LHC or ILC via higgsino pair production but is certainly testable at higher energy hadron colliders with 30–100 TeV
Aesthetic Mechanisms of Dehumanization in the Prison Industrial Complex
Get PDFMy dissertation argues that the aesthetic conditions of incarceration seriously impact the experiences and treatment of incarcerated people in the United States. In Chapter I, I argue that humans have aesthetic needs which are integral to our wellbeing. Aesthetic needs include the need to enact aesthetic agency as well as the need to have aesthetically fulfilling experiences. When aesthetic agency is severely constrained, as it is in many carceral contexts, aesthetic harm may occur. Chapter II discusses the relationship between aesthetic harm and dehumanization. Aesthetic harm can be understood both as a form of and a sign of dehumanization. As a form, aesthetic harm can be understood as a method of dehumanization which utilizes aesthetics. As a sign, aesthetic harm refers to the phenomenon where prior exposure to dehumanization reinforces future subjection to aesthetic harm. Thus, there is a feedback loop occurring with aesthetic harm: prior exposure to aesthetic harms as a form of dehumanization increases the likelihood of continued exposure to mistreatment, both aesthetic harm and other types of harm. Chapter III explores examples within two categories of aesthetic harms: those associated with the conditions of the environment incarcerated people are subjected to, and those which come from the objects they encounter in the prison. Designing goods and spaces with the goal of better managing and controlling a population and saving costs is often at odds with building a space which provides support for the needs of the incarcerated, something which is essential for rehabilitation. Chapter IV examines acts of aesthetic resistance. Aesthetic resistance involves attempting to regain or retain agency related to one’s everyday aesthetic experiences and creative expression. These acts work to resist dehumanization by countering the aesthetic harm which creates and exacerbates dehumanization in the carceral context
Improved Learning Through Neural Component Search
Get PDFDeep learning models contain many different hyper-parameters that need to be
tuned prior to training. These hyper-parameters greatly influence the quality of
the final model. Historically, most attention and research has been performed
on tuning the architecture. However, with the advent of automated machine
learning and the success from neural architecture search, automated methods have
been successfully applied to other components of neural networks, challenging
the very inspiration of the classical methodologies. In this work, automated
methods, through the use of evolutionary algorithms, were applied to the learning
components of a neural network: the loss function, optimizer, learning rate
schedule, and output activation function of computer vision models with the goal
of finding drop-in replacements for standard components. I expand upon previous
research in each of these respective domains through the proposal of new search
spaces, surrogate functions, genetic algorithms, and better found components. In
the end, multiple loss functions, optimizers, learning rate schedules, and output
activation functions, all evolved from scratch, were found to be able to outperform
cross-entropy, Adam, one cycle cosine decay, and softmax on the CIFAR datasets
OUTLIER EXPLANATIONS FOR DATA STREAMS: OUTLYING ATTRIBUTES AND ROOT CAUSE FACTORS OF OUTLIERS
No full textData streams, which are continuous sequences of timestamped data points, necessitate real-time monitoring due to their time-sensitive nature. They have special characteristics such as the notion of infinity (continuous arrival of data points and unbounded volume of data) and concept drift. Detecting outliers which are data points significantly different from the rest of the data in the dataset, is crucial in many data stream applications. For example, in network security and credit card transaction monitoring, real-time detection of outliers is vital, as these outliers often signify potential threats. However, investigating detected outliers usually requires significant time and effort from the users. Therefore, providing real-time outlier explanations is equally important, as it enables users to gain insights and shorten their investigation time. When an outlier is detected as a multidimensional object, the investigation time of the outlier is equivalent to the number of outlier attributes. Hence providing an outlier explanation in the form of the set of attributes responsible for the outlier abnormality (also known as the outlying attributes) is necessary. In some applications such as cloud monitoring, expert users spend considerable time and effort to investigate the root cause factors of an outlier detected in the front-end service. The investigation of the root cause factors involves examining the front-service all the way to the backend services. Providing an outlier explanation in the form of root cause factors of the outlier is important to minimize user effort in identifying the reasons for their occurrences. There exist techniques that discover outlying attributes and root causes factors of outliers for data streams. However, they do not simultaneously address the characteristics of data streams, especially for those involving the notion of infinity and concept drift.
This dissertation proposes two outlier explanation algorithms, EXOS and Ocular, for discovering outlying attributes and root cause factors of outliers, respectively. EXOS is designed for discovering outlying attributes of multi-dimensional outliers in data streams. Unlike other existing techniques, EXOS leverages cross-correlations among data streams, accommodates varying data stream schemas and arrival rates, and effectively addresses challenges related to the unbounded volume of data and concept drift. The algorithm provides real-time explanations based on the local context of the outlier, derived from a time-based tumbling window. Ocular is an algorithm designed to identify root cause factors of point outliers in continuous real-time data streams. It utilizes a user-provided normal causal graph, which depicts the causal relationships or dependencies between variables in a system. When the value of a variable at a particular timestamp is detected as an outlier, Ocular employs this causal graph to identify the variables responsible for the anomalous value of the target variable. The algorithm simultaneously addresses inherent characteristics of data streams: the notion of time, the notion of infinity, and concept drift.
Extensive theoretical and empirical analyses have been conducted to evaluate the performance of EXOS and Ocular using both real and synthetic datasets. The evaluation results show that, on average, EXOS achieves a 45.6% better F1 Score and is 7.3 times lower in explanation time compared to existing outlying attribute algorithms. Additionally, Ocular outperforms current root cause identification algorithms by 170% in F1 Score on average, while maintaining comparable or lower explanation times
DISTRIBUTED MATRIX ANALYSIS AND COMPUTATION OVER NETWORKS
Get PDFThis thesis introduces a continuous-time distributed algorithm designed to address a range of matrix analysis and computation problems in networked systems. Focusing initially on the Local-Equation Local-Variable (LELV) problem, the algorithm enables nodes within the network to collaboratively tackle six specific challenges. These include computing least-squares solutions to linear equations, determining the minimum-norm least-squares solution, detecting solution existence, computing the Moore-Penrose inverse of a matrix and identifying full column or row rank matrices.
The algorithm, functioning as an affine, networked dynamical system, demonstrates global exponential convergence, supported by an explicit lower bound on its convergence rate. Furthermore, it offers deterministic guarantees for some problems while ensuring convergence with probability one for others.
Extending the scope to include the Local-Equation Global-Variable (LEGV) problem, this thesis provides preliminary analysis, including equilibrium point analysis and simulation of the algorithm to demonstrate convergence. While minimal in-depth exploration was conducted, these initial insights highlight the algorithm’s potential applicability in addressing LEGV challenges within distributed environments.
Overall, this thesis contributes a novel continuous-time distributed algorithm with
significant implications for matrix computation in networked systems. Through rigorous theoretical analysis and initial exploration, it lays the groundwork for further research and practical applications in distributed computing settings
Perceived Problem Behaviors in Pre-Kindergarten: The Role of Teacher-Child Racial Match and Teacher-Child Relationship
Get PDFIn 2020, during the height of the COVID-19 pandemic, the national enrollment rate for pre-k children ages 3 to 4 was 40% (a drop from 54% in 2019; National Center for Education Statistics, 2022). This is partly due to fewer children in the 0 – 5 age group (23.4 million of 72.8 million children in the United States in 2020; U.S. Census Bureau, 2020). With the increase in enrollment of multiracial pre-k children (National Center for Education Statistics, 2022) and the general projected population increase of children of color (U.S. Census Bureau, 2020), diversity in pre-k classrooms is expected to dominate. By contrast, 79% of United States school teachers are White, and that percentage increases to 90% at predominantly White schools (National Center for Education Statistics, 2022). Many of these teachers have little preservice training related to issues of diversity, equity, and inclusion (Brown et al., 2016; Miller & Mikulec, 2014) which can leave them unprepared once in the field. To counter this deficit, teachers can intentionally deploy critical self-reflection to examine if they are engaging in equitable teaching practices. Unaddressed deficits can lead to misunderstandings, microaggressions, and implicit biases that can manifest in discriminatory behaviors, especially when there are extreme power imbalances between a teacher and a 3- or 4-year-old developing child of color. The disproportionality of suspensions of Black pre-k children is 2.5 times greater than other racial groups of pre-k children and 35.8% of pre-k children expelled and 45% of pre-k children suspended were Black boys outranking other pre-k children (Civil Rights Data Collection, 2021). Black girls made up 20% of pre-k girl enrollment but 54% of pre-k girls suspended (Civil Rights Data Collection, 2021). Expulsion and suspension are stressful and negative experiences for children, their families, and their providers, and can set off a negative trajectory. Research indicates that expulsion and suspension early in a child's trajectory predicts expulsion and suspension later in life. Children who are expelled or suspended from school are as much as ten times more likely to drop out of high school, experience academic failure, hold negative school attitudes, and face incarceration than those who are not.
First, this study proposes a new theory, critical race attachment theory (Benabdallah, 2020), to expand teachers' knowledge base by synthesizing critical race theory (Crenshaw et al., 1995) and attachment theory (Bowlby, 1969) in addition to incorporating a self-reflective component. The convergence of these two theories into one comprehensive framework, critical race attachment theory, broadens, strengthens, and deepens the understanding of developmental and cultural intersectionality, supports authentic cultural competency, and encourages the scientific teacher method. Integrating a self-reflective model emphasizing culturally appropriate development as a guiding educational tool to understand the importance of teachers as attachment figures and evidence-based authentic data collection plays an essential role in developing a teacher's pedagogical approach. Second, skewed or unexamined racialized and gendered biases affect perceptions of a child's behavioral performance in the classroom, particularly when the child is Black. To explore this widely accepted phenomenon, a secondary data analysis was conducted on a large quantitative data set from the combined National Center for Early Development and Learning (NCEDL) Multi-State Study of Pre-Kindergarten and the State-Wide Early Education Programs (SWEEP) study (Early et al. 2013). The current study had a sample size of N = 470 children. This sample focused exclusively on Black and White teacher-child pairings and problem behavior scores measured by the Teacher-Child Rating Scale (TCRS; Hightower et al., 1986). In the current study, findings were mixed. Whereas race was not associated with problem behavior, gender was associated with problem behavior. Boys were rated much higher by their teachers, on average, than girls. Thus, critical race attachment theory would be useful in increasing teachers' pedagogical cultural-development knowledge and self-reflective practices. Third, this dissertation provided practical strategies for implementing the critical race attachment theory as a real-world model for secure teacher-child attachment regardless of race and gender.
This dissertation is formatted as three manuscripts. The first manuscript emphasizes theory and provides a critical lens to view culturally appropriate teacher-child attachment in early childhood settings using critical race attachment theory. The second manuscript focuses on empirical evidence and utilizes a large dataset to perform secondary analysis utilizing multiple regression to support practical strategies guided by theory. The third manuscript has a practical emphasis and provides the practitioner with useful strategies based on theory and evidence
EVALUATION OF MACHINE LEARNING ALGORITHM ACCURACY IN ENERGY EXPLORATION USING PHYSICS DERIVED SYNTHETIC MODELS.
No full textOil and gas exploration and production struggles with meeting energy needs while minimizing environmental impact. Amidst artificial intelligence (AI) and machine learning (ML) to expedite energy exploration and minimize risks associated with energy investments through enhanced predictive accuracy. The application of AI/ML holds promise in expediting the identification of geological facies linked to reservoir rock and delineating seismic faults at seismic scale or generated 3D velocity models based on image and not forward modeling. However, before fully harnessing AI/ML capabilities, it's imperative to rigorously assess its ability to address these challenges with precision and confidence. This study aims to evaluate how machine learning can augment geoscience practices by enhancing accuracy, managing a multitude of 3D seismic attributes, and overcoming limitations in real-world interpretation.
The work is divided into four chapters, each exploring different algorithms and methodologies. Chapter 2 uses generative adversarial networks (GANs) with geostatistical seismic inversion to characterize the Ray Reef gas storage area and generate alternative reservoir models. Chapter 3 compares unsupervised machine learning techniques with established geophysical methods for capturing reservoir facies and fluid effects using seismic attributes, validated on synthetic and field data. Chapter 4 employs unsupervised machine learning to classify lithologies and estimate fault probability in a large offshore seismic dataset, evaluating lateral seal risks. Finally, Chapter 5 delves into explainable machine learning (LIME) to understand uncertainties in 3D subsurface water saturation modeling, comparing random forest and recurrent neural network predictions.
The combined findings of this research underscore the significant potential of machine learning techniques to enhance various aspects of oil and gas exploration and production workflows. By leveraging the power of generative adversarial networks, the study demonstrates their effectiveness in generating alternative reservoir models, providing valuable insights for reservoir characterization and risk assessment. Furthermore, the application of unsupervised learning algorithms showcases their capability in identifying reservoir facies and fluid effects from seismic data, offering a powerful tool for reducing interpretation uncertainties. The integration of explainable machine learning techniques, such as LIME, sheds light on the decision-making process of complex models, enabling a better understanding of subsurface water saturation modeling and associated uncertainties. Collectively, the results highlight the transformative impact of machine learning in geoscience, paving the way for more efficient and informed decision-making processes in oil and gas operations while mitigating environmental risks.
Throughout the dissertation, the importance of integrating domain knowledge, rock physics principles, and synthetic data generation is emphasized to validate machine learning models and ensure physically meaningful results. The work highlights the potential and limitations of these techniques, advocating for a balanced approach that combines data-driven methods with domain understanding in geoscience applications
DIURNAL CHANGES IN SURFACE URBAN HEAT ISLANDS AND THEIR RELATIONSHIPS WITH URBAN LAND COVER
Get PDFUrbanization, a prominent phenomenon since the industrial revolution, has led to significant changes in land use and land cover (LULC), resulting in the Urban Heat Island (UHI) effect. This effect, characterized by higher temperatures in urban areas compared to their rural surroundings, is influenced by the use of heat-absorbing materials, human activities, and urban geometry. The resulting high temperature affects urban residents with implications for public health, greater energy consumption, and change in local climatology. Understanding the spatial variations in land surface temperature (LST) and the thermal effects of heterogeneous urban forms is essential for developing mitigation strategies to enhance the urban thermal environment.
A modern method introduced in the past decade in urban climatology is the concept of the local climate zone (LCZ). This approach disaggregates the heterogeneity in urban settings and classifies surfaces into various built and natural land cover types, which in turn helps to interpret the patterns of LST in the local surrounding area. Previous studies have shown that different urban structures, such as residential areas, commercial zones, and industrial regions, exhibit varying surface temperature patterns. In developing countries, urban planning often results in heterogeneous urban landscapes with mixed land uses, further complicating the UHI dynamics. This study investigates the relationship between various LCZs and their LSTs in the cities of Ahmedabad and Surat in the state of Gujarat in India. The LCZ map used in this study was created from a fusion of two data sets: 1) a freely available 100m global LCZ map, and 2) a 30m land use land cover (LULC) map developed using Landsat-8 data.
A few studies have explored the diurnal variations of LST over urban areas, mainly due to the revisit cycles of polar orbiting satellite (e.g., Landsat Series, Terra, Aqua). However, in this study, we leverage LST data from The National Aeronautics and Space Administration's (NASA) latest Ecosystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) that provides LST at 70m resolution at different diurnal periods. Then we studied how LST changes within different LCZs at various times of day to understand the thermal effects of urban landscapes.
The results reveal distinct seasonal and diurnal variations in LST. During winter, Surat exhibited consistently higher temperatures than Ahmedabad, with differences averaging around 5°C. In summer, temperatures rise significantly in both cities, peaking at 40°C in Surat and 45°C in Ahmedabad. Several LST hotspots were identified, particularly in industrial areas and densely built-up zones. For instance, an international airport in Ahmedabad showed temperatures up to 52°C on summer afternoons, significantly higher than the surrounding areas.
Distinct thermal patterns were observed across LCZs during different diurnal periods. Compact midrise areas were the hottest, while open lowrise areas were cooler during both seasons and across all diurnal cycles in both cities. Large lowrise areas were typically hotter than compact lowrise areas in the mornings and afternoons. However, large lowrise areas cooled faster, making them cooler than compact lowrise areas during the evenings and nights. In both cities, compact midrise and compact lowrise areas exhibited the highest temperatures in the evening and nighttime.
The study also analyzed the diurnal variations in LST gradients from the center to the edges of the city, revealing that during summer mornings, there was a negative correlation between LST and distance from the city center. This correlation weakened in the afternoon but strengthened again in the evenings and nights, particularly in Ahmedabad. Both cities showed stronger negative correlations in the evenings and nights, indicating a pronounced UHI effect while in the afternoon there was no correlation as the solar radiation was heating the surface uniformly.
This study emphasizes the value of using the new ECOSTRESS LST products for analyzing diurnal thermal variations with fine spatial resolution. These findings offer an understanding of the impact of various urban structures on local climate aiding city planners and developers in implementing informed heat mitigation strategies. However, the use of ECOSTRESS data comes with limitations, such as the day and night LST data not being from the same day, causing inaccuracies and uncertainties. Additionally, cloud cover can result in missing grid cells, and variations in viewing angles due to the ISS's orbit can affect the accuracy and consistency of LST measurements
Exploring U.S. Climate Attitudes 2023 SPEER Survey Findings
Get PDFWe present findings from a comprehensive survey of climate-related attitudes across the United States in 2023. Using an online sample of 2,188 U.S. adults, we examined correlates of climate-related attitudes. Regional analyses reveal regional variations in climate attitudes, with the West South-Central region showing consistently high levels of concern across most measures, while the East South-Central region demonstrates the lowest levels of concern. The Pacific region exhibits a nuanced pattern, ranking
high in risk perception and concern for future generations, but lower in personal worry and anticipated harm. Linear regression analysis for climate belief identifies several significant predictors including political affiliation and orientation, religious beliefs, education, and urbanicity. Conservatives, Republicans, evangelical Christians, and rural residents are more likely to express skepticism about climate change, while higher education levels correlate with stronger climate change beliefs. Additionally, we assessed public trust in climate scientists using a feeling thermometer scale, finding generally high trust levels with notable regional and demographic variations. This inaugural survey establishes a baseline for tracking changes in climate attitudes over time and provides valuable insights for tailoring climate communication strategies and policy approaches across different segments of the U.S. population.N