Treasures @ UT Dallas
Not a member yet
7697 research outputs found
Sort by
Renal Clearable Gold Nanoparticles with Different Surface Chemistries : Fundamentals and Applications
No full textThe past decades have witnessed significant progress in the field of nanomedicines, where
numerous nanomaterials have been designed to address challenges in disease detection and
therapy. However, the slow clinical translation of nanomedicines highlights the importance of
fundamental studies of nano-bio interactions and in vivo nanoparticle (NP) transport because only
with a solid understanding of their in vivo behaviors can nanomaterials be applied to successfully
address challenges in disease diagnosis and treatment. According to recent U.S. Food and Drug
Administration guidance on using nanomaterials in human drug products, nanomaterials are
required to be eliminated from the body rapidly to prevent long-term toxicity, which emphasizes
the importance of the renal clearance pathway of NPs. There are many factors that could influence
the renal clearance of NPs, such as size, density, surface charge, and surface chemistry. This
dissertation aims to focus on unraveling how the surface chemistry fundamentally affects the in
vivo transport of renal clearable gold nanoparticles (AuNPs) in normal and injured kidneys. In
Chapter 1, I discuss the underlying reasons why glutathione can serve as an effective capping agent
for enhancing renal clearance of ultrasmall AuNPs in the native physiological environment and report the critical length of the surface ligands for efficient renal clearance. In Chapter 2, I report
a thorough investigation of the in vivo transport of glutathione-coated AuNPs (GS-AuNPs) in a
mouse model with cisplatin-induced nephrotoxicity and reveal the fundamental understanding of
how drug-induced kidney injuries of different degrees could impact the transport and retention of
GS-AuNPs in the different kidney compartments. In Chapter 3, I discuss an unexpected discovery
that an antifouling ligand can slow down the renal clearance of AuNPs while still making the NPs
high resistance to serum protein and remain renally clearable. Interestingly, we find that this
AuNPs can chemically stimulate neurons at both the in vitro and in vivo levels, which has never
been observed before
Prosodic Correlates of Doubt and Trust: Production, Perception, and Listener Experience
No full textAffective prosody is the use of suprasegmental speech features (fundamental frequency, pauses,
speech rate, and intensity variation) to convey information about the talker’s emotions or attitude.
Attitudinal prosody, which conveys a talker’s stance or judgment about something or someone,
has (until recently) rarely been investigated. This study takes up the topic of attitudinal prosody by
exploring how American (US) English speakers convey two specific attitudes: doubt (disbelief)
and trust (belief). These attitudes are considered because of their significance in social sensitive
situations like the investigative interview between law enforcement personnel and victim or
conversations between doctors and patients. In a pilot speech production and perception study
(Chapter 2), native speakers of US English successfully distinguished both doubting and trusting
prosody from neutral prosody.
In a speech production experiment (Chapter 3), untrained talkers’ productions of doubting and
trusting prosody are elicited using two syntactic structures and in different imagined social
scenarios. Social information, (e.g., the identity of the listener), may alter how talkers use prosody,
but, until the present study, this prediction had not been tested. Production experiment results
suggest that prosodic features indicating “trust” may be more evident when the talker is in a
position of authority relative to the listener.
An issue likely contributing to the paucity of attitudinal prosody research is an unbounded set of
attitudes that must be conveyed with a limited number of prosodic features. In Chapter 4, this issue
was addressed in a perception experiment, in which listeners described the talker’s attitude after
listening to recordings from the production experiment by choosing as many terms as apply from
a list of 20 affect labels. Listeners rarely chose the terms “doubting” and “trusting”. Rather,
listeners distinguished doubt-, neutral-, and trust-intending tokens with constellations of terms that
aligned with “doubting” and “trusting” on dimensions like valence and arousal. Listeners’
groupings of labels were also correlated with acoustic features like utterance duration and initial
fundamental frequency.
The sensitivity of perception of doubting and trusting prosody is addressed in Chapter 5, with an
experiment where listeners were presented with gradient manipulations of doubting and trusting
prosody in synthesized tokens of the word “okay.” Listener perceptions of the talker’s attitude
were found to be mediated not merely by the prosody in a trial, but by prosody on previous trials
from the same talker. Perception of talker attitude was also sensitive to manipulations of
perceived talker role (“speaker”, “doctor”, or “police officer”) and whether the listener was
experiencing symptoms of post-traumatic stress. These variables were of interest because of their
relevance to the situations of the investigative interview and conversations between doctors and
patients. More broadly, understanding how interpretations of prosody change with relationships,
conversational context, and listeners’ past experiences can provide guidance for future attitudinal
prosody studies. The results presented throughout this dissertation suggest that attitudinal
prosody research might be advanced by considering how prosodic variables are interpreted
across contexts and map to a minimal set of affective dimensions, rather attempting to define a
set of prosodically distinct attitude
Religious Ritual in Domestic Decoration on the Roman Bay of Naples
No full textThis thesis investigates depictions of religious rituals found in domestic contexts. The first
chapter focuses on the depictions of sacrifice to the Lares and Genius on the lararia of six
distinct houses: The House of Julius Polybius, the House of Sutoria Primigenia, Villa 6 at
Terzigno, a newly discovered house in Regio IX, a taberna in Regio VII and a lararium with an
unknown findspot. The second chapter examines the cult of Dionysus and the ritual of initiation,
an event possibly recreated in room 5 of the Villa of the Mysteries. The third centers on the cult
of Isis and the rituals surrounding religious celebrations, specifically ones recreated in two panels
discovered in Herculaneum during the early days of Bourbon excavations. After a thorough
study of the visual representations, this study expands to consider the role and meaning of
religious ritual in multiple religions on the Roman Bay of Naples. Each ritual represented has its
own history and meaning across various cultural groups. These rituals offered an opportunity for
an individual to interact with a god and reinforced their connection to a specific belief system.
The depiction of religious ritual within a domestic context celebrates a family’s beliefs and
documents their adherence to the expectations of specific religious customs
Calculated Re-vision: Kennedy, Johnson and African American Views of Their Civil Rights Legacies
No full textSince the 1960s, there has been broad scholarly interest in the civil rights legacies of President
John Kennedy and his successor, President Lyndon Johnson. Examinations have emerged
from a wide range of disciplines, but it has been almost thirty years since the only book-length
study of this subject appeared. Mark Stern’s Calculating Visions: Kennedy, Johnson and Civil
Rights (1992) argued that neither Kennedy nor Johnson was particularly committed to civil rights
when they joined forces on the Democratic Party ticket in 1960, and both were political
moderates who eventually succumbed to the pressure applied by civil rights idealists. Stern’s
analysis, with its heavy reliance on presidential administration records and former staff
members’ memoirs and interviews, overlooked a key question. If both Kennedy and Johnson
were viewed as political moderates, why have they been understood so differently by the
African American communities most impacted by their civil rights policies?
This dissertation addresses that question by focusing on African American responses to the civil
rights strategies of Kennedy and Johnson. Mining African American oral histories, memoirs,
letters, speeches, telegrams, essays, material culture, newspaper and magazine articles, polling
data, song lyrics, visual art, and filmed portrayals, it traces how perceptions about these leaders’
civil rights records developed in the 1960s and continue to circulate today. The resulting
analysis highlights the trajectory by which Kennedy emerged as a civil rights hero for black
Americans while Johnson became a figure of relative contempt and mistrust. It explores the
ways African Americans aligned themselves with Kennedy’s memory over Johnson’s reality as a
form of black countermemory, drawing an invisible dividing line between the time many believed
integrated, government-led, non-violent social change was possible, and when many no longer
maintained that hope.
A central component of this research deals with the manner by which John Kennedy has been
mourned as a civil rights martyr within the black community. African Americans have imbued
Kennedy’s image with a meaning that serves their ongoing, everyday struggle for racial equality,
affording him a privileged presence in their homes. The portraits of Kennedy in black
households operated as hidden transcripts that communicated his unique value to future
generations. Despite Lyndon Johnson’s effort to enact historic civil rights legislation that many
African Americans acknowledge went further than anything John Kennedy likely would have
supported, Johnson never achieved sustained personal affection from black voters. Although
African Americans were vital to Johnson’s landslide reelection victory in 1964, they continued to
believe that his support for civil rights was motivated by political self-interest rather than a
sincere commitment to racial equality. Representations of Johnson in recent civil rights films
perpetuate a narrow view of him as a racist manipulator. The passage of fifty years warrants a
calculated re-visioning of these two presidents’ civil rights legacies, and how they have been
perceived by African Americans in their own time and since. This effort challenges long-held
perceptions of the roles John Kennedy and Lyndon Johnson retain in both the Civil Rights
Movement and in the African American imagination
IOT Integration, Adversarial Attacks, And Threat Explanations In Provenance-based Intrusion Detection Systems
No full textSystem provenance analysis has become the predominant approach for defending against
sophisticated attackers. System provenance analysis captures causal and informational flow
dependencies by correlating telemetry data across key system resources such as processes,
files, and network sockets. These dependencies are efficiently represented as system prove-
nance graphs, which are directed, heterogeneous, and multi-attributed. These system prove-
nance graphs can be used by Provenance-based Intrusion Detection Systems (PIDSs) to train
adaptive behavioral Machine Learning (ML) models for intrusion detection tasks. PIDSs can
effectively thwart Advanced Persistent Threat (APT) actors and Fileless Malware writers
since they can measure the program behavioral deviations. Graph Neural Networks (GNNs)
are the de-facto standard for learning from graphs. Consequently, GNN-based PIDS can
detect zero-day and mimicry attacks by measuring deviations in program behavior.
Despite their undeniable advantages, modern PIDSs still face several open problems: (1)
current system provenance analysis techniques are designed primarily for resource-rich en-
vironments, leaving IoT ecosystems vulnerable; (2) the resilience of PIDS against dedicated
adversaries have not been fully examined; (3) GNN-based PIDS operate as black-box models,
lacking transparency in their detection decisions.
This dissertation addresses these three key challenges in system provenance analysis: ex-
tending provenance analysis to IoT environments, improving robustness against adversarial
attacks, and enhancing the explainability of GNN-based PIDS.
First, we introduce ProvIoT, a federated edge-cloud security framework that brings PIDSs
to resource-constrained IoT devices. ProvIoT leverages federated learning to minimize
network and computational overhead while maintaining high accuracy in detecting stealthy
attacks, even in diverse real-world environments.
Next, we present ProvNinja, an adversarial testing framework designed to evaluate the
robustness of PIDSs against realistic evasive attacks. ProvNinja generates adversarial
attack variants that closely mimic benign system behaviors, allowing it to effectively test
the resilience of State-of-The-Art (SOTA) PIDSs. Our experiments reveal vulnerabilities in
current security models, leading to reduced detection rates in realistic attack scenarios.
Finally, we develop ProvExplainer, an explainability framework for GNN-based PIDSs to
provide interpretable, security-focused explanations. ProvExplainer projects the GNN’s
decision boundaries onto the interpretable surrogate model’s feature space (e.g., discrimi-
native subgraph patterns). By integrating with SOTA GNN explainers, ProvExplainer
improves both precision and recall in explaining stealthy attacks (i.e., APTs campaigns and
Fileless malware) detections, offering a transparent and verifiable tool for security operations.
Together, these contributions offer scalable, robust, and explainable security solutions for
increasingly interconnected and vulnerable digital infrastructure
Lighthouse: A Compendium of the Visual Development Process
No full textLighthouse: A Compendium of the Visual Development Process delineates an approach of
developing a narrative for an animated film beyond the initial visual exploration. This paper
establishes the current methodology of commercial productions within the customary phase of
pre-production, where the project is grounded in visual research and preliminary conceptual
design based on a screenplay or narrative brief. The author will demonstrate the strata of pre-
production stages utilizing an original project created in 2018 to provide a broad overview of the
process of taking the written word through the stages of research and design to create a visual
framework for the remaining production phases to build upon. There will not be specific
disclosure of tools used in the creation of this project unless they have direct relevance to the
process. The focus remains primarily on the methodology of the execution of the thesis. Drawing
from three decades of industry experience the author will also touch upon a possible landscape
where the design process continues throughout the production pipeline and how the methodology
may be valuable in assisting the evolution the project narrative
Tropical Diseases and Transnational Migration in Indonesia
No full textThe existing literature on tropical diseases focuses on the effect of migration on the transmission
of tropical diseases, such as Dengue and malaria. However, the literature neglects the possible
effect of tropical diseases on migration, a research gap that this study is contributing to close.
Using panel data from the Podes dataset on the village level allows this study to observe the
influence of outbreaks of malaria and dengue on transnational outmigration. This study finds a
positive connection between dengue outbreaks and migration, with female migrants driving the
effect. Furthermore, this study suggests that malaria decreases migration in rural islands like
Kalimantan. Nevertheless, the study suffers from endogeneity and uncertainties about the
magnitude of infections during an outbreak, making the results unreliable to generalize
The Impact of Leadership Styles of Public Managers on Effectiveness, Job Satisfaction, and Employee Reward Systems
No full textAs local governments strive to attract effective public managers and foster high morale among
their managers and staff, this mixed-methods dissertation focuses on leadership styles and their
relationships with leader effectiveness, leader job satisfaction and employee rewards. In the
spring of 2021, top-level public managers (N=211) from cities across Texas completed a
Multifactor Leadership Questionnaire (MLQ) to investigate the relationships between public
managers’ self-perceived leadership styles—and leadership behaviors—and their self-perceived
levels of leader effectiveness and leadership satisfaction. Twenty-eight of the MLQ respondents
were interviewed in the summer of 2022 to determine the job aspects that cause the leaders’
satisfaction and to learn whether leadership style impacts the types of employee rewards that
public managers deem most effective. The quantitative phase of this study found positive
correlations between transformational leadership and self-perceived levels of effectiveness and
leader satisfaction. Two behaviors—idealized influence (attributes) and contingent reward
behavior—were also associated with these leadership outcomes. In the qualitative phase of the
study, all the transformational public managers interviewed emphasized sources of intrinsic
motivation as the aspects of their jobs that give them the most satisfaction. Being able to support
their staff successfully and making a difference in their communities were especially valued by
the participants. The interviews also yielded insights about the rewards given by
transformational and contingent reward transactional leaders. Both groups of leaders prioritize
motivators over hygiene factors, non-external regulation over external regulation, and non-
monetary rewards over monetary rewards
Neurobiological Correlates of Rhythm and Syntax Processing
No full textLanguage comprehension requires temporal analysis of a sequence of words that incorporates an
overarching rhythmic and syntactic structure. Emerging evidence indicates that musical rhythm
skills are associated with grammatical competence, suggesting common neurobiological traits for
rhythm and syntax. This thesis aimed to elucidate genetic and neural correlates of rhythm and
syntax processing implicated for temporal processing in the sensorimotor system. The first study
examined the potential influence of three dopamine genotypes, DRD1, DRD2, and COMT, on
rhythm and grammar skills. Results showed that rhythm performance measured via spontaneous
and synchronized finger tapping as well as rhythm discrimination was positively associated with
language performance on grammaticality judgment and syntactic comprehension of spoken
sentences. Among the three genes, only DRD1 gene influenced rhythm skills. In addition, the
DRD1 polymorphism accounted for grammar skills indirectly via the association between
rhythm and grammar skills. The second study examined the role of sensorimotor beta oscillation
as a common neural correlate for rhythm and syntax. In this study, the degree of modulation in
beta oscillatory power during finger tapping or passive listening to a metronome beat was used
as a neural index of rhythm-related beta activity to predict beta oscillatory dynamics involved in
grammaticality judgment or sentence comprehension. Results on sentence comprehension
showed that beta activity during beat tapping was associated with beta activity related to
syntactic demand, i.e., an increase in beta power for object-relative compared to subject-relative
clause sentences. By contrast, there was no significant association regarding grammaticality
judgment. In a second experiment of this study, beta activity during beat listening was associated
with beta activity related to syntactic surprisal, i.e., a decrease in beta power for the object-
relative clause condition when it was less frequent than in the first experiment. These results
suggest that there are dissociable components in the relationship between neural traits of rhythm
and syntax processing. Taken together, the findings of this thesis suggest the sensorimotor
system as a neuroanatomical substrate underlying the connection between rhythm and syntax
Understanding and Improving the Efficiency of Machine Learning Software: Model, Data, and Program-level Safeguards
No full textMachine learning (ML) software has become integral to various aspects of daily life, leveraging
complex models such as deep neural networks (DNNs) that entail significant computational
costs, especially during inference. This poses a challenge for the deployment of ML software
on resource-limited embedded devices and raises environmental concerns due to high energy
consumption. Addressing these challenges requires improving the efficiency of the ML
software.
ML software consists of three core components: data, model, and program. This dissertation
investigates efficiency optimization for these components. Existing research primarily focuses
on model and program optimization but overlooks the critical role of data. Additionally, the
robustness of model-level optimizations and the compatibility of program-level optimizations
with model-level ones remain underexplored.
This dissertation aims to address these gaps. At the data level, it examines how training
data impacts ML software efficiency and proposes techniques to mitigate efficiency vulnerabilities introduced by adversarial data. At the model level, it explores the robustness of
dynamic neural networks (DyNNs) to efficiency degradation and presents methods to enhance
their inference efficiency. At the program level, it introduces a novel approach to bridge
program-level and model-level optimizations, ensuring comprehensive efficiency improvements.
Moreover, it also analyze the model leakage in the model acceleration process.
The contributions of this dissertation are threefold:
Data-level: This research evaluates the impact of training data on ML model efficiency,
identifying efficiency backdoor vulnerabilities in DyNNs and proposing strategies to defend
against them. Model-level: It examines computational efficiency vulnerabilities in DyNN
architectures, developing tools like NMTSloth and DeepPerform to test and mitigate these
vulnerabilities. Program-level: This dissertation introduces a program rewriting approach,
DyCL, designed to adapt existing DL compilers for DyNNs, significantly enhancing inference
speed. Additionally, this dissertation proposes an automatic method, NNReverse, which
infers the semantics of the optimized binary program to reconstruct the DNN model from
the compiled program, thereby quantifying model leakage risks.
Overall, this dissertation provides a comprehensive framework for optimizing ML software
efficiency, integrating data, model, and program-level approaches