62588 research outputs found
Sort by
Explorations of Racism-Related Stress, Mindfulness, and Psychological Health
No full textPerceived racism is associated with adverse mental and physiological health outcomes for marginalized racial and ethnic groups, including symptoms of anxiety, depression, and stress. Mindfulness, including the practice of compassion, may be one helpful coping strategy to mitigate the effects of racism-related stress. Mindfulness has been shown to moderate the effects of anxiety, depression, and distress. Although research on mindfulness and stress has increased in recent years, mindfulness has been rooted in religious, secular, and spiritual traditions among people of color and known to have connections with numerous multicultural traditions and religions long before inclusive in Western research. A dearth of research exists in the mindfulness literature around the relationships between racism, mindfulness, and psychological health outcomes among people of color. As such, additional research is warranted to understand stress and coping mechanisms used by people of color to combat the effects of racism. The current study used a cross-sectional research design to examine the relationships between recent experiences of racism and psychological outcomes (e.g., anxiety, depression, stress, hope) for people of color, and explored the relationships of recent experiences of racism with psychological outcomes for people engaging in various levels of mindfulness. Results indicated there were significant relationships between recent racism experiences and psychological distress as well as significant relationships between recent racism experiences and hope. Results also suggested that mindfulness served as a buffer between recent racism experiences and psychological distress, such that people of color with higher levels of mindfulness experienced less psychological distress in response to recent racism experiences compared with people of color with lower levels of mindfulness. Lastly, results suggested that people of color with higher levels of mindfulness may maintain relatively stable levels of hope regardless of their recent experiences with racism. Understanding of relationships may add further knowledge in the stress and coping literature to support the psychological well-being of people of color impacted by racism.</p
Daddy Issues: Facilitators and Barriers to Gender Transformative Father-Son Communication about Gender Based Violence, An Exploratory Study
No full textCisgender boys/men perpetrate the vast majority of gender based violence (GBV) globally, including intimate partner violence and sexual violence. GBV has many devastating consequences including physical injury, reproductive issues, depression and post-traumatic stress disorder, inability to work and lost wages, and homicide. It is important to create effective prevention and intervention programming to target GBV among boys/men. Emerging evidence exists that targeted programming for boys/men that incorporates a gender transformative approach may be more likely to shift men’s gender and violence related attitudes and behaviors more effectively. Gender transformative programming explicitly focuses on critically examining gender-related norms, expectations, attitudes and behaviors. Father-son GBV communication that involves gender transformative tenets may be a new avenue of GBV prevention. This study aimed to be the first to examine how fathers who have discussed GBV with their adolescent sons may use gender transformative features to conduct these discussions. Additionally, to inform future GBV prevention programming efforts with fathers and sons, the study investigated facilitators and barriers that fathers perceive for conducting these gender transformative GBV conversations with their sons. Results indicated that several fathers in this sample used gender transformative traits in their father-son GBV conversations, and most fathers saw utility in a gender transformative approach in father-son GBV conversations. Implications, recommendations and future directions based on insights from this study are included.</p
Secondary Special Education Teachers’ Knowledge of Word Reading: A National Survey
No full text{"value":"Secondary special education teachers require foundational reading content knowledge to adequately address the needs of students with word reading difficulty. Yet, measures of secondary teachers’ knowledge are lacking. I developed the Word Reading Knowledge Assessment for Secondary Teachers (WRKAST) to assess teachers’ knowledge of phonemic awareness, phonics, syllabification, and morphology content. I recruited a nationally representative sample of 285 Grades 6-12 special education teachers to complete a 50-item online survey that included the WRKAST measure and questions pertaining to instructional settings and practices. On average, teachers answered approximately 18.6 of the 28 WRKAST items correctly (SD = 4.9). I conducted confirmatory factor analysis (CFA) and Rasch analysis to examine the psychometric properties of the WRKAST. CFA results suggested an underlying factor structure comprised of four subscales: phonemic awareness (PA), phonics, syllabification, and morphology. The data also fit a unidimensional Rasch model . PA items were significantly more difficult than morphology, syllabification, and phonics items. Together, years of experience, percentage of time spent teaching word reading, enhanced training in reading, and grade level band were not significant predictors of WRKAST subscale scores. These results have several implications for research, practice, and policy. First, the WRKAST needs further optimization, including development of additional items that target the upper range of the word reading knowledge variable. Second, teacher-specific variables that better predict word reading content knowledge should be identified. Finally, results indicate that secondary special education teachers may need more support with developing specialized knowledge of the structure of language, a requisite for effective instruction.
","attr0":"abstract"
Formal Methods for Multi-Robot Systems: Scalable and Robust Planning with Temporal Logic and Partial Satisfaction
No full textThis dissertation addresses the challenges posed by the increasing complexity of multirobot systems by proposing a formal methods framework that ensures scalable, robust, and customizable mission planning. The primary contribution is the integration of high-level temporal logic specifications, specifically Signal Temporal Logic (STL) and its weighted extension (wSTL +), with Mixed-Integer Linear Programming (MILP) formulations. This integration enables the automatic synthesis of controllers that meet various mission objectives. Novel MILP encodings have been developed to enhance the expressivity and computational efficiency of STL and wSTL, allowing their application in large-scale scenarios. The framework is also extended to model complex multiagent dynamics, including swarm behaviors, modular robot reconfigurations, heterogeneous team coordination, and resource-constrained logistics. Furthermore, this work introduces a systematic approach to handling infeasible specifications through partial satisfaction, ensuring that mission-critical objectives are prioritized even when some constraints cannot be fully met. Overall, this dissertation advances the state of the art in multi-robot planning by combining formal temporal logic reasoning with optimization-based control synthesis, providing a principled and practical solution for real-world uncertainties and specification infeasibility.</p
Experimental Study of Wall-Induced Translation of a Rotating Particle in a Shear-Thinning Fluid
No full text{"value":"The interactions of swimmers with boundaries play an important role in biological and technological systems. Various swimmers use their interaction with nearby surfaces for propulsion. Translation–rotation coupling is one of the ways to do this, where rotation close to a boundary is directly translated into directed translation. These interactions are further influenced by the properties of the surrounding fluids. Fluids in many biological environments are non-Newtonian. They often show shear-thinning rheology, in which an increase in shear rate results in a decrease in viscosity. These characteristics influence swimmers\u27 motions, particularly the translation-rotation coupling close to boundaries. Understanding the influence of those factors on swimmer behavior near walls is essential for further steps in both research and applications.In this study, we experimentally investigated the wall-induced translation of a rotating sphere in shear-thinning fluids across a range of moderate Reynolds numbers. Using a magnetically actuated sphere driven by Helmholtz coils, we examined the swimmer\u27s behavior in three shear-thinning and one Newtonian fluid at three distinct distances from a rigid boundary. Each condition was repeated three times, and the motion was quantified through image analysis techniques.
The results reveal that both inertial and rheological effects contribute to the observed propulsion behavior, with directional reversal occurring only at close boundary proximity and in strongly shear-thinning fluids. As wall distance increases, reversal disappears and the swimmer\u27s motion becomes more uniform, suggesting that neither inertia nor viscosity contrast is sufficient to drive distinct propulsion without wall-induced asymmetry. These findings complement prior numerical predictions and offer experimental insight into swimmer behavior in complex fluids, with potential relevance for micromachine design and control in real-world environments.
","attr0":"abstract"
Functionalizing Solvent-Cast Chitosan/Polycaprolactone Membranes for Artificial Corneal Replacement
No full text{"value":"Globally, millions of people suffer partial or complete vision loss from a diseased or damaged cornea. Cadaver corneal transplants are the primary treatment however the limited supply of cornea donor tissue does not meet the increasing demand. Corneal tissue engineering has demonstrated significant progress with supporting the regeneration or replacement of damaged corneal tissue through the use of natural and/or synthetic polymers in combination with cornealcells and bioactive molecules. However, challenges in developing tissue-engineered corneas include insufficient mechanical strength for sutureability, low optical transparency, and poor cell-material interactions for tissue integration. To address these challenges, membranes were fabricated via solvent-casting with crosslinking agents to enhance mechanical stability and surface functionalized with peptide-PCL and polymer conjugates to influence local cell behavior. In this
work, membranes were crosslinked using natural crosslinking agents such as genipin, peptide-PCL conjugates with the amino acid sequence CYGGGRGDS were synthesized, and PEG-PCL conjugates were used. Membranes were fabricated with genipin, pre-functionalized with RGDS(biotin)-PCL or PEG-PCL conjugated in a single-step functionalization strategy without post-fabrication modification. Membranes crosslinked with the highest concentration of genipin enhanced mechanical stability while exhibiting a higher transmittance compared to human eye bank cornea. In single-functionalized membranes, increasing the peptide-PCL conjugate concentration resulted in a significant increase in NIH3T3 fibroblast adhesion compared to PEG-PCL functionalized membranes. Introduction of RGDS(biotin)-PCL and PEG-PCL conjugates slightly affected optical properties but remained within the range of human eye bank cornea. Functionalized membranes cultured with cells retained transparency. Dual-functionalized membranes with RGDS(biotin)-PCL and PEG-PCL conjugates within a single membrane demonstrated spatial organization of conjugate functionalization. Cells preferentially attached to RGDS(biotin)-PCL compared to PEG-PCL, showing how tailored spatial organization can be used to control cell behavior within a single membrane. This work introduces a solvent-cast based platform for creating membranes with tunable mechanical and biochemical properties for TE applications like corneal replacement.","attr0":"abstract"
Inexact and Stochastic Large-Scale Nonlinear Optimization
No full textNonlinear optimization algorithms often use iterative methods. Typically, starting from an initial point, an algorithm computes a step, determines a step size, and updates the solution estimate using the step and step size. The type of iterative method that is employed should depend on the size of the problem, namely, whether it is small or large. In this thesis, we focus on methods for solving large-scale problems. An optimization problem can be large in different responses; for example, the number of variables may be large and/or a dataset that defines the problem functions may be large. This latter type is common in areas such as machine learning. This thesis addresses nonlinear optimization algorithm design for solving large-scale problems of both of the aforementioned types.First, we design a trust-region method for solving nonconvex unconstrained problems when the number of variables is large and solving the trust-region subproblem exactly is intractable. Another goal of the algorithm is to maintain the optimal worst-case complexity of that is achieved by a trust-region method based on exact subproblem solutions. To achieve this, we propose some sufficient conditions that the inexact subproblem solutions must satisfy to maintain the optimal complexity, as well as a computationally efficient Krylov method for obtaining inexact solutions that satisfy these sufficient conditions. Second, when the dataset describing an optimization problem is large, evaluating the gradients, objective, etc., becomes expensive. This setting applies to optimization problems in machine learning, such as those minimizing expected risk and empirical risk. We address such settings that also involve nonlinear constraints on the design variables. Specifically, we consider cases where only minibatch (i.e., stochastic) gradient of the objective are accessible, but the constraint, and the constraint derivative are available. We design two groups of algorithms—Sequential Quadratic Programming (SQP) and Interior Point Methods (IPM)—both with guarantees to converge to a KKT solution under reasonable assumptions. We also apply our SQP method to solve physics-informed machine learning problems, demonstrating improved performance over unconstrained approaches, such as penalty methods.</p
Compression-Aided Privacy and Inferential Separation in Machine Learning
No full textThe rapid proliferation of Internet of Things (IoT) devices and the demand for real-time data processing have raised significant concerns about data privacy in machine learning applications. This dissertation addresses these challenges through two key approaches: inferential separation and compression-aided privacy.In inferential separation, we develop methodologies to protect sensitive inferences drawn from high-rate data streams, without compromising data utility. This includes a theoretically grounded framework for protecting sensitive inferences in IoT systems, as well as Decoct-Net, a deep learning-based model designed to sanitize sensitive attributes without compromising non-sensitive information.
In the domain of compression-aided privacy, we explore techniques that remove sensitive information from computational models while maintaining their utility. This includes Spectral-DP, a spectral domain perturbation method that enhances the utility of differentially private learning through spectral filtering, and two theoretically rigorous approaches-Randomized Quantization with SGD (RQP-SGD) and Gaussian Sampling Quantization for Federated Learning (GSQ-FL)—which focus on achieving privacy and communication efficiency in resource-limited environments.
By combining theoretical insights with empirical validation, this dissertation demonstrates how sensitive information can be effectively removed from data and models. The proposed techniques provide significant advancements in privacy-preserving machine learning, particularly in IoT and edge computing environments, without sacrificing model performance.</p
Stress, Attributions, and Affiliate Stigma: Profiles of Parents of Young Children At-Risk for ADHD
No full textYoung children with attention-deficit/hyperactivity disorder (ADHD) experience impairment associated with negative outcomes; parenting practices can have a major impact on the trajectory of ADHD symptomology and related impairment. Negative attributions, affiliate stigma, and parental feelings of stress can deleteriously impact parenting practices, increasing the risk for negative parent and child outcomes. The identification of parents with negative attributions, affiliate stigma, and higher levels of parenting stress in early childhood can allow psychologists to recognize families who are at-risk for problematic outcomes and implement more comprehensive and individualized treatment. The purpose of this study was to examine if there are distinct subgroups of parents of preschoolers at-risk for ADHD based on parenting stress, attributions, and affiliate stigma, and if child- (i.e., oppositional and aggressive behaviors) and parent- (i.e., parental ADHD, education status) related variables are associated with profile membership. Four latent profiles were obtained that differentiated parents based on severity of stress, attributions, and affiliate stigma. Child oppositional and aggressive behaviors and parental ADHD symptomology were significantly associated with parent profile memberships with more challenges. Findings indicate the need for screening methods to identify parents who are experiencing high levels of stress, attributions, and affiliate stigma as they may require more intensive intervention of longer duration. Results also highlight the potential importance of treatment individualization as well as the need to further examine possible pre-treatment intervention response determinants, profile membership over time, and the impact of intervention on profile membership
When the Woman Screams: Female Horror Screams as a Reclamation of Space, Agency, and Monstrosity
No full textIn 1984, Linda Williams penned the landmark essay “When the Woman Looks,” a critique of how the female gaze in horror films reflects the woman’s cultural status of subordination and her association with monstrosity. This dissertation addresses what comes after that moment of identification: the female scream. When the Woman Screams challenges the idea of screaming as an inherently feminine act fueled by terror by suggesting that the scream acts as a disruptor to the male gaze, not least by shifting audience focus to (and in potential alignment with) the female survivor. Functionally, the scream is as transgressive as the monster’s violence. When a woman screams in horror, she is occupying a space of competing cultural categories of acceptable behavior. These vocalizations, whether stemming from fear, anger, grief, or a combination of emotions, resist cultural frameworks that have historically equated “good” womanhood with silence. The substantial narrative power that underlies female horror screams arises as a disruptor of this silence-- highlighting inequitable power structures-- and it is in this spirit of disruption that When the Woman Screams reimagines the format of the dissertation. Inspired by the nontraditional dissertation work of Anna Williams and Amanda Visconti, When the Woman Screams is a multimodal project comprising four parts. Housed on a dedicated website, these parts may be consumed independently or in conversation with one another. The podcasts represent my interest in the blending of cultural history and narrative storytelling. They are designed to center the orality of screams within horror films and to connect those screams to a broader historical context. Here, mode is important for capturing the mechanics of the scream (tone, roughness, pitch) and separating the scream from its performance, such that the screams exist as auditory markers of on-going American cultural conversations. Conversely, the video essay exploring silent screams highlights the performance aspect of screaming and demonstrates the ways in which gender norms are being reflected or challenged. Because these screams lack verbal cues, audiences must rely on the visuals of the performance to contextualize meaning. Film, then, provides a mode for elucidating how perceptions of the bodily movement of women are not fixed but culturally fluid. A third part of this project - dedicated blog entries providing close readings of select films - democratizes scholarly discourse by making complex argumentation publicly accessible to a global audience. Rounding out the project are its citations. By placing scholarly literature and popular literature in conversation with one another, these citations serve as a way of dismantling perceptions of what constitutes legitimate data. </p