18525 research outputs found
Sort by
Comparing Self- and Parent Report: Sensory Features and Language In Autism
No full textPSY 4999: Honors Thesis
Faculty Mentor: Mark Wallace, PhD
Background: No prior studies have compared self- and parent-reported questionnaires for
autistic and non-autistic children. The relationship between self- and parent-report on core and related features of autism such as language also remains unclear. The purpose of this study was to examine differences in self- and parent-reported patterns of hyper- and hypo-responsivity in autistic and non-autistic children age 7-17 years. Differences in how both reports map onto expressive and receptive language were also examined between groups. Participants included seven autistic individuals and five non-autistic individuals. Self-reported hypo- and hyper- sensory constructs were assessed with the Glasgow Sensory Questionnaire (GSQ). Parent- reported hypo- and hyper- sensory constructs were assessed with the Sensory Experiences Questionnaire (SEQ). Expressive and receptive language were assessed using the Clinical Evaluation of Language Fundamentals-Fourth Edition (CELF-4). Kruskal Wallis Tests were conducted to determine if GSQ and SEQ hyper- and hypo- sensory constructs differed by diagnostic group. Correlations were also conducted to determine if there were associations between both reports and expressive and receptive communication skills.
Results: There was no significant difference between GSQ and SEQ scores or between the autistic and non-autistic group in either the hyper- and hypo-responsivity domains. However, there was a more robust relationship between SEQ hyper-responsivity and expressive language in both groups.
Conclusion: This study extends prior work to examine how self-report differs from parent-report when assessing sensory responsivities linked to language in older autistic children.Thesis completed in fulfillment of the requirements of the Honors Program in Psychological Science
Defending Model Extraction Attacks with Probabilistic Isolation
No full textDeep Neural Networks (DNNs) are an essential intellectual property for companies providing Machine Learning as a Service (MLaaS) due to their exceptional capabilities in domains such as text generation, generative graphics, image recognition, and robotics. Due to their often-proprietary nature, the training datasets and model architectures are lucrative targets for malicious actors looking to violate user privacy, gain a competitive market advantage, or generate adversarial examples. This thesis presents Jigsaw, a deep learning defense framework that partitions computation of a DNN to different nodes in a datacenter to provide a moving target defense against model extraction attacks. The defense is based on the insight that partitioning DNN computation can improve latency, that modern large language models require operator parallelism to scale, and that side-channel attacks are already computationally demanding on a singular target, let alone multiple. Jigsaw employs a moving target defense strategy that uses probabilistic pseudo-isolation to decrease the time value of information necessary to launch a model extraction attack. By trading the strong privacy guarantees of trusted execution environments or data-oblivious computation for probabilistic pseudo-isolation, we can secure larger models with less performance sacrifices. We show that with Jigsaw, there is an increase in the resources and time required to successfully complete an attack, which allows time to detect malicious behavior through well-studied anomaly detection systems. To bound the performance impact within the large random partitioning space of a DNN, we use a genetic algorithm to identify candidate partition plans that maintain reasonably strong performance while also providing a sufficient reduction in the attacker's window of opportunity. By balancing security with performance, Jigsaw aims to provide MLaaS providers with a practical security solution that scales to current large models and anticipates the needs of tomorrow's technologies
Design and Development of a Temporal Event Annotation Tool to Support Collaborative Debriefing for Nurse Training
No full textWithin healthcare education, simulation-based training is a popular and successful approach that focuses on providing students with hands-on experiential learning in a controlled, safe, and repeatable environment. Debriefing is considered to be one of the most important components of the simulation process. Debriefing is a feedback approach that encourages learners to critically analyze their performance, identify areas for improvement, and construct new knowledge through directed feedback, reflection, and discussion immediately following an experiential simulation event. However, there are several practical and psychological barriers to effective, best-practice aligned debriefing. Motivated by these challenges, this dissertation aims to reduce barriers to effective simulation debriefing in nursing education by designing and developing a supportive technology tool called PULSE.
Through a user-centered design process that engaged nursing instructors and administrators, PULSE was developed to support instructors in addressing several challenges that they face in guiding nursing students through debriefing processes. PULSE is organized around a timeline of “moments”, which are a set of events of pedagogical interest and associated comments marked by instructors and observing students while watching a live simulation. After the simulation concludes, the PULSE system aggregates these marked moments, performs text-based learning analytics on the comments, and presents all this information alongside recorded video of the simulation on an organized debriefing dashboard designed to facilitate discussion among the group. This dissertation presents the user-centered design process of PULSE and a set of field studies that were conducted to refine the dashboard visualizations and evaluate its efficacy. A mixed-methods analysis of the study data revealed that, under the right conditions, PULSE can support improved structure and depth of debriefing conversations, primarily by providing students with increased opportunities to share their input during instructor-guided debriefing discussions. Through the design, development, and analysis of PULSE, this dissertation provides an evidence-centered approach to support a more comprehensive understanding of nurse training scenarios that can lead to deeper reflection among students. In addition, this research contributes new evidence and insights to ongoing discussions in the simulation-based training community about best practices for debriefing and the role of observing students in the debriefing process
Ultra-Low-Overhead Arbitrary-Waveform Generation as a Circuit Macro: Augmenting the Characterization of Radiation-Induced Transient Effects in Highly Scaled Integrated Circuits
No full textArbitrary waveform generators (AWGs) are not typically feasible as subsystems on integrated circuits due to their size and complexity, but they are versatile circuits that are broadly useful. The purpose of this work is to show that by prioritizing minimal overhead and designing for targeted performance, as necessary for the application, it is possible to create a reusable on-chip AWG circuit macro in a small form factor. To support this claim, details and results are provided for a proof-of-concept implementation in a 45nm partially depleted silicon-on-insulator process. The presented design is able to achieve a small size by eliminating the complicated calibration and filtering circuitry commonly used in contemporary designs, instead relying on intrinsic accuracy of the base circuits. The reliability and accuracy of the AWG are driven by careful design down to the layout level, including the development of a variant of the traditional common-centroid layout technique called distributed-centroid layouts (DCL), which addresses the importance of bias circuitry in mitigating process-induced mismatch. A custom simulation workflow was developed to investigate the effectiveness of this technique at the circuit level as compared to other designs from the literature. The proof-of-concept circuit was designed to improve the characterization of radiation-induced transient effects in highly scaled integrated circuits by providing built-in self-test and hardware-emulation capabilities to a custom photocurrent measurement circuit (PMC). Details of this specific application are explored in detail, along with experimental measurements made using flash x-ray and pulsed laser sources. Alternative AWG designs that might benefit a broader application space beyond radiation effects are also provided
Causal Approaches to Quantifying the Role of Engagement in Studies of Mobile Health Interventions
No full textRecent technological advancement has resulted in the proliferation of interactive text message-based interventions to support medication adherence in patients managing chronic illnesses. Several recent clinical trials have identified these interventions as a strategy to improve outcomes, particularly when used in combination with other interventions. In these settings, patient engagement with these text messages may drive a portion of the intervention’s effects on key outcomes. Such trials typically include a control arm with no opportunity to engage with text messages. Nevertheless, the relationship between engagement and outcomes may be subject to unmeasured confounding. Quantifying treatment effects using engagement as a post-randomization variable is therefore challenging. In this dissertation, we develop approaches to handle these challenges and provide researchers with principled tools to understand the role of engagement with mobile health interventions. Our first focus involves methods to estimate and bound functional local average treatment effects (i.e. an effect of treatment at theoretical levels of engagement under the intervention), when the exclusion restriction cannot reasonably be assumed. We investigate these methods cross-sectionally and longitudinally in a regression-based framework, and derive closed-form sandwich variance estimators for key contrasts of interest. We further show that this method accommodates multiple pathways from treatment to outcome, and consider how operationalizing engagement over time can affect these approaches. Our next focus involves direct investigation of engagement as a mediator, suitable for the setting in which we believe key common causes of engagement and the outcome have been measured. The first fundamental goal of this aim is to delineate (and interpret) the mediation effects that are applicable to studies of mobile health interventions under strong access monotonicity, and the second is to formalize the assumptions under which they can be identified. We propose using a parametric g-computation based approach to estimating key effects, and evaluate finite-sample properties through simulation studies. We illustrate the utility of our proposed methods through application to a recent clinical trial of patients with type 2 diabetes that showed significant overall effects on key psychosocial outcome measures
Understanding how DNA methylation patterns at enhancers record cellular histories
No full textDNA methylation is an epigenetic modification that is essential for proper multi-lineage cellular differentiation, though the connection between patterned regions of hypomethylation and phenotype is poorly understood. Over 80% of CpG sites in the genome are methylated, while continuous genomic regions featuring low methylation levels form hypomethylated regions (HMRs). While promoter HMRs are largely consistent across cell types and cell states, subsets of non-coding HMRs are cell-specific and enriched for tissue specific regulatory elements. Our objective was to better understand the implications of patterns of non-coding HMRs both within and between cell types. In this thesis, we systematically dissect HMR patterns across human cell types and tissues representing a diversity of developmental stages, including embryonic, fetal, and adult tissues. Unsupervised hierarchical clustering applied to the methylomes across 11 cell types and tissues, representing >100,000 HMRs, revealed cluster groups that define distinct developmental stages, cell lineages, and individual cell types. Tracking HMRs through pseudo time course representing the hematopoietic lineage revealed that most HMRs (~ 70-75%) in differentiated cells are established at earlier developmental stages and accumulate through cell specification, suggesting a model of hierarchical developmental establishment. HMRs acquired through differentiation frequently (~ 35%) establish near existing HMRs (≤ 6kb), resulting in the formation of HMR clusters associated with stronger enhancer activity. We employed a SNP-based partitioned heritability analysis using GWAS summary statistics across diverse traits and clinical lab values. This analysis indicated that HMRs enrich for cell-relevant trait heritability, increasing with both sequential developmental specificity and HMR clustering. Moreover, cell-specific HMRs enrich for heritability for cell-relevant traits at levels above other enhancer annotations. Our analyses emphasize the power of HMR subsets to predict cell phenotypes; they also highlight DNA methylation as a unique epigenetic mark that provides genetically distinct records of a cell’s history through development
Barriers to Entry For Women-and Minority-Owned Businesses in Government Contracting
No full textWomen- and Minority-Businesses (WMB) encounter persistent barriers to participate in government construction contracts in the United States. Despite longstanding policies aimed at increasing WMB participation, various factors, including bonding requirements, subcontracting programs, and payment practices tend to exclude WMBs from fully participating in government markets. This dissertation investigates these exclusionary effects at both the Federal and State levels, offering evidence that should help inform future policy decisions.
Chapter I examines the impact of bonding requirements on WMB participation in government construction projects. Analyzing a 2010 change in the minimum dollar threshold for requiring bonding requirements, I find that WMB participation increased in contracts for which the requirement was removed. Moreover, the study finds larger effects for small WMB participation. These findings provide initial empirical evidence that surety bond requirements within the construction industry act as a barrier to entry for WMBs.
Chapter II evaluates the effects of subcontracting requirements on procurement prices. I examine both California and Iowa’s use of the Department of Transportation’s Disadvantaged Enterprise Program (DBE). For California, I find that bid prices increased by 13%, in response to a large change in subcontracting goals. In Iowa, I find that bid prices were sensitive to variation in contract-level goals but not to changes in overall state goals.
Chapter III investigates WMB participation California’s Department of Transportation contracts, comparing state and federal projects. Exploiting the fact that federally funded projects use federal procurement law, and state funded projects follow state law, I find that the WMB share of bids and contracts awarded in federal projects is much higher than for state projects. I suggest that differences in payment practices and the existence of the DBE program for federal contracts is associated with higher WMB participation rates.
Overall, WMB representation in government contracts is still far lower than their presence in the economy would suggest. Identifying barriers and effective program evaluation is an essential step in improving diversity and equity in government markets. This research contributes valuable evidence to inform policy interventions designed to enhance WMB participation in government contracting
THREE PAPERS ON POSTSECONDARY ACCESS AND SUCCESS IN TENNESSEE
No full textThe purpose of this dissertation is to better understand postsecondary access and success in the context of Tennessee. Postsecondary access and success are important policy issues to explore as participation in the U.S. labor market increasingly requires a postsecondary credential. This three-paper dissertation examines these issues by focusing on two aspects of postsecondary access and success: rural residents, who may face unique, structural barriers to postsecondary access, and the role of public assistance programs in providing essential basic needs assistance to students pursuing college degrees. In the first, co-authored paper, Dr. Adela Soliz and I describe the landscape of rural postsecondary access in Tennessee by exploring regional variation in rural college-going using state administrative data. In the second paper, I build on this work by estimating the effect of a supplemental college counseling program in rural high schools in Tennessee. In the last paper, I move beyond college access to focus on a particular college persistence issue: basic needs insecurity. I use interview and focus group data from students and administrators from three community colleges in Tennessee to describe students’ participation in three public assistance programs: the Supplemental Nutrition Assistance Program (SNAP), Temporary Assistance for Needy Families (TANF), and Women, Infants, and Children (WIC). Together, these papers contribute to existing literature on postsecondary access and success by highlighting the unique contexts that frame rural residents’ college-going, evaluating the impact of an intervention for increasing rural residents’ postsecondary access, and describing how students’ engagement with public assistance programs is important for facilitating their college persistence
Effects of Class-Wide Function-Related Intervention Teams (CW-FIT) on Middle School Teachers' Praise and Reprimand Frequency
No full textThere is a large body of empirical research showing the effectiveness of teacher praise as an intervention to reduce unwanted classroom behaviors and increase student engagement. Despite this evidence, research shows that teachers do not naturally provide praise at high rates. Over the past 60 years, researchers have tested numerous methods to improve rates of teacher praise including prompting, visual performance feedback, and peer coaching. In our study of 15 teachers, we investigate the effects of Class-Wide Function-Related Intervention Teams (CW-FIT) on middle school teachers' rates of praise and reprimands. Using a regression analysis, results showed that CW-FIT increased teachers' frequency of praise compared to teachers not using the CW-FIT intervention. Future directions and implications for using tier 1 behavior management systems to increase levels of teacher praise are discussed