Treasures @ UT Dallas
Not a member yet
7697 research outputs found
Sort by
Non-resilience of Resilient Distributed Consensus in Multi-agent Systems
No full textThis thesis explores the resilient distributed consensus in networks that lack the necessary
structural robustness to achieve consensus in the presence of malicious agents. While exist-
ing solutions provide robustness conditions for consensus among normal agents, they fail to
evaluate network performance comprehensively when the graph’s robustness is insufficient.
To address this limitation, we introduce the concept of non-convergent nodes, representing
agents unable to attain consensus with any arbitrary agent due to malicious agents in the
network. This notion allows us to classify graphs based on their robustness levels and assess
partial performance. This study initially establishes the (r, s)-robustness of commonly en-
countered graphs, such as complete, complete bipartite, 1-D distance, and circulant graphs.
Our approach facilitates easier identification of robustness and enables us to gain insights
into the behavior of non-convergent nodes. By understanding the dynamics of these non-
convergent nodes, we can establish more relaxed conditions for converging subgraphs, which
are the subgraphs that are guaranteed to converge. This knowledge enhances our under-
standing of resilient algorithms and their behavior in practical scenarios. Furthermore, we
present graphs with given robustness levels, including (F + 1, 1), (F, F ), and (F + 1, F ) ro-
bustness, and determine the maximal number of non-convergent nodes associated with each
graph. This quantification of non-resilience sheds light on the impact of graph robustness
on the network’s ability to achieve consensus. Surprisingly, we find that graphs with the
same structural robustness may exhibit varying degrees of non-resilience, leading to different
network performance outcomes. Through numerical evaluations, we demonstrate that our
approach provides a comprehensive resilience perspective beyond the conventional binary
view of success or failure in the face of malicious agents. By quantifying network perfor-
mance under sub-optimal robustness conditions and identifying converging subgraphs, our
study opens up new possibilities for designing more resilient consensus algorithms
Guided Subset Selection Based Unified Active Learning Framework: Formulations, Algorithms and Applications
No full textDeep learning has been successful in a wide variety of domains, ranging from face recognition
to self-driving cars. The success of deep learning algorithms is due to large amounts of labeled
data, which is easily available in today’s digital world. However, training deep models using
large datasets comes with high compute costs and labeling costs. Moreover, large datasets are
often naturally plagued with data imperfections like class imbalance, OOD, and redundancy.
Training machine learning models using such datasets leads to biased models obtained even
after incurring expensive costs. Moreover, these models underperform on rare yet critical
scenarios, which can be catastrophic. As one can imagine, one or more of these problems can
occur at any point during the development of machine learning models. Hence, there exists
a need for a wholistic framework that can serve as a one-stop solution for data-efficiency,
model-efficiency and reducing data imperfections. In this dissertation, we focus on designing
a Guided Active Learning framework that can serve this purpose. In particular, we propose
four phases of the Guided Active Learning framework: 1) Seed Set Selection, 2) Discovery,
3) Targeting and, 4) Filtering. This framework is designed to be modular, since different
teams can find themselves in requiring to optimize for different phases in this framework.
The Seed Set Selection phase focuses on finding an initial set that represents the larger
dataset, such that the majority of the information and semantics of the dataset are covered.
The Discovery phase focuses on finding unknown instances that do not exist in the current
labeled dataset or were potentially missed during the exploration phase. The Targeting phase
aims to mitigate data imbalance by finding data points that are semantically similar to rare
classes or slices. The Filtering phase focuses on avoiding out-of-distribution and redundant
data points from being selected.
We provide algorithms and mathematical formulations for executing these phases in several
applications. Particularly, we demonstrate the effectiveness of the Guided Active Learning
framework on a wide range of real-world domains including autonomous driving, medical
imaging and automated speech recognition. We hope that the Guided Active Learning
framework will help practitioners navigate data better and improve the performance of their
downstream machine learning models
Suburbacology
No full textSuburbacology is made up of two games: Neko Hanshoku and Animal House. Both games
address human interaction with suburban wildlife and how humans are often the cause of any
problems that arise. Neko Hanshoku looks at feral and free-roaming cats that have become an
invasive species fueled by people. Animal House portrays a world where humans and animals
have switched roles to show how violent pest removal is. Both games are considered
"unwinnable" because there is no happy ending; this makes the message more impactful and
forces the player to come to terms with the reality of the situation. Suburbacology makes the
player see the direct impact of their actions without giving them the option to fix the problem.
Research for this project shed light on how inhumane treatment of animals is considered
"humane" because details are often left out of the extermination description. Humans decided the
hierarchy of life importance—where humans are on top—and because of this many people feel
certain animal lives are more important than others, or not important at all. Most animals only
exist in the suburbs because of the free food and shelter provided by humans, but once animals
that are not desirable start appearing, they are the pests. There are no easy solutions to these
problems, and the answers will look different for each suburb, but there are answers available.
Co-existing peacefully with animals is an option if people take the time to understand them a bit
better. The goal of this thesis is to make people consider alternative ways of handling animals in
suburban areas, rather than the inhumane methods commonly practiced under the guise of pest
control
Gallium Oxide Semiconductor MOS Capacitors With Atomic Layer Deposited High-K Dielectrics
No full textBeta-Gallium Oxide (β-Ga2O3) has garnered significant interest as a semiconductor for high-power
and optical devices because of its wide bandgap and its availability in wafer form, in addition to
supporting a wide range of epitaxial materials with various dopants. A variety of dielectric
materials are available for creating MOS devices using β-Ga2O3 substrates. A suitable dielectric
must have a high dielectric constant and a sufficient bandgap offset to prevent Fowler-Nordheim
or direct tunneling of electrons. Al2O3, SiO2, and HfO2 are potential candidates due to relatively
high band offsets with β-Ga2O3 and conceivable ease of integration. This work is composed of
several studies which investigate the effects of surface treatments, such as chemical cleaning,
plasma etching, or heat treatments of the β-Ga2O3/ dielectric interface, on MOS capacitor
performance using capacitance-voltage and current-voltage analysis techniques. The purpose of
this research is to contribute to the overall development of β-Ga2O3 technology especially in the
areas of dielectric interfaces and oxide reliability
Tilting at Translation—the Misery and Splendor of Transduction: Moving the Meaning of García Lorca’s Mariana Pineda From Spanish to English and From Page to Stage
No full textThis creative dissertation centers on a varied set of translation processes of Federico García
Lorca’s Mariana Pineda: first from Castilian to English and then from script to performance.
The dissertation contains a series of six essays, a playscript (in translation), the link to a
performance of a version of that playscript, and appended collateral materials of the production
process. The series of essays explores the practical and theoretical limits of translation as a
process, both specific to the work with Mariana Pineda and abstracted from it. The playscript is
meant for theater-makers of all kinds—and the performance, as the end result of a translation
process, may be useful to others who choose to produce the playscript
Examining Suicidal Ideation as a Mediator Between Stress and Aggressive Police Misconduct
No full textPrior research links suicidal behaviors and aggressive forms of police misconduct to stress.
However, no study examines whether suicidal cognition, specifically suicidal ideation,
influences verbal and physical police misconduct. Psychological and sociological theories
suggest that self- and other-directed violence are related, but social and cultural constraints
determine the outcome. First, this dissertation examines whether suicidal ideation mediates the
relationship between stress and aggressive police misconduct, specifically verbal abuse and
excessive force, and the relationship between negative affect and aggressive police misconduct.
Second, this dissertation investigates the effect negative affect (depression, burnout, and anger)
has on aggressive police misconduct and whether anger exerts the greatest influence. A series of
multinomial logistic regressions are conducted with mediation analysis following the guidelines
set by Baron and Kenny (1986) and utilizing Karlson, Holm, and Breen’s (2011) decomposition
method. Primary and supplemental analyses indicate that stress and negative affect’s relationship
with verbally abusive police misconduct is partially mediated by suicidal ideation. Mediation
does not occur when observing the excessive use of force exclusively and in the presence of all
forms of negative affect. The analysis also indicates that, individually, all forms of negative
affect increase aggressive police misconduct. Whether separate or paired with the other forms of
negative affect, anger is consistently the greatest emotional influence on aggressive police
misconduct. The findings of this dissertation suggest that suicidal ideation may underlie verbal
aggression exhibited by some law enforcement officers. This supports the theoretical perspective
that suicidal behavior and external aggression are linked, but social constraints prohibit certain
displays of aggression. Furthermore, it demonstrates that the processes under Agnew’s General
Strain Theory [GST] (1992) may be dynamic and influence multiple behavioral outcomes.
Concerning GST, this dissertation further supports that anger is a key factor in perpetuating
violent behavior. Given the implications of these findings and that organizational stress is
consistently cited as the most influential form of stress as seen in prior research (Amaranto et al.,
2003; Bishopp et al., 2016, 2019; Violanti et al., 2019), law enforcement agencies should
consider promoting officer well-being. Solutions directed at organizational approaches to stress
reduction are likely to make the greatest gains in reducing behavior resulting from stress
detrimental to the organization and the individual. Ignoring organizational issues will likely
promote further issues in policing
Multiple Sclerosis and Healthy Aging: a Comparative Analysis of Structure and Function
No full textThe relationship between healthy, normal aging and the neurodegenerative disease Multiple
Sclerosis (MS) is a subject of ongoing debate. Parallels between the two include vascular and
metabolic decline in addition to cognitive decline. In particular, processing speed declines are
seen in both aging and MS. In aging, previous work has suggested that processing speed decline
is thought to be the basis for all age-related cognitive decline. Whereas neurovascular and
neurovascular effects of MS and aging have previously been studied separately, they have not
been studied in parallel. In addition, the relationships between these effects and MS- and age-
related cognitive decline have not previously been investigated. In the studies presented here, I
explore the relationships between neurovascular and neurometabolic function in aging and in MS
as well as their effects on cognitive and brain volume using multiple regression analysis. I found
that cerebrovascular reactivity is associated with cognitive decline in aging, but not in MS,
despite significantly lower arterial reactivity in MS. I also found that task-evoked neurovascular
and neurometabolic dynamics are associated with brain volume in MS, but not in aging. Lastly, I
do not find any relationships between processing speed and white matter integrity in any group.
These results indicate that processing speed decline is a diffuse phenomenon that can arise from
the dysfunction of many different systems. It is also indicative of divergent effects of vascular
pathology and signaling dysfunction on MS patients and older adults. This work shows that
despite similarities in pathological processes, vascular and metabolic associations with cognitive
decline and brain volume differ between MS and aging
Modeling and Sensitivity Analysis for Trace Gas Sensors
No full textTrace gas sensors can detect very low concentrations of gases such as methane and sulfur dioxide.
An important class of trace gas sensors are quartz-enhanced photoacoustic spectroscopy (QEPAS)
sensors, which employ a quartz tuning fork (QTF) and modulated laser to detect trace gases. Existing models of QEPAS sensors employ one-way coupling from the fluid to the structure that requires
prior experimental measurements of the damping of the QTF due to its motion in the viscous fluid.
We study an improved two-way coupled model that is based on a Helmholtz system of thermo-
visco-acoustic equations in the fluid, together with a system of equations for the temperature and
the displacement of the structure. These two subsystems are coupled across the fluid-structure
interface via several conditions. With this model, the user specifies the geometry of the structure
and the viscous and thermal parameters of the fluid, and the model outputs an effective damping
parameter and a signal strength that is proportional to the concentration of the trace gas. We derive
analytic solutions of the two-way coupled model in the special case that the QTF is replaced by an
annular structure. This simplification of the geometry allows the pressure, temperature of the fluid,
and the displacement of the structure to be expressed in terms of Bessel functions. These solutions
show reasonable agreement between the one-way and two-way coupled models at higher ambient pressures. However, at low ambient pressure the one-way coupled model does not adequately
capture thermo-viscous effects. For the two-way coupled model, excellent agreement is obtained
between the analytical results and simulations performed using a finite element formulation of the
model.
Computational models for trace gas sensors involve a large number of parameters. If one wants
to quantify uncertainty of the output quantities of interest (for example, pressure, temperature or
displacement of the tuning fork tines) then one must estimate statistical distributions that describe
these quantities. However, statistical studies require that one runs the physical simulator for hundreds or thousands of different input parameter values. This process is computationally prohibitive
unless one can first identify which parameters influence the model output. We use the active sub-
space method to identify a subset of parameters that is influential for the output. The application of
the active subspace method to the pressure-temperature subsystem in the special case of cylindrical
symmetry identifies one influential parameter for the fluid temperature and three for the pressure
from the five dimensional parameter space. Similarly, for the two-way coupled model with annular
geometry the active subspace method reduces the 13 dimensional parameter space to a 5 dimensional subspace by identifying 4 influential parameters for the fluid temperature and 5 influential
parameters for the remaining quantities of interest. Finally, for the one-way coupled model with
tuning fork geometry, the active subspace method identifies 5 influential parameters for the fluid
pressure, temperature, and displacement of the QTF reducing a 10 dimensional parameter space to
a 5 dimensional subspace. These results also show an excellent agreement with the results obtained
using kernel density estimation and a simple sensitivity study
Effective Video Solutions for Earth Science Education
No full textWith one in seven people worldwide regularly using digital communication platforms such as
YouTube, an empirical understanding of how geoscience is communicated and educate with
videos is crucial. This understanding may be beneficial for reaching audiences who are less
likely to engage with geoscience-related content in mass media or other traditional channels.
Also, although there has been significant research about how to design better and use videos for
geoscience educational and communication purposes, the different goals and contents of
communication via video lead to various instructional video designs. With increasingly easy and
inexpensive video production and dissemination, more and more geoscience educational videos
on varied Earth Science topics are becoming available. However, how to best create and utilize
geoscience videos to formally and informally teach diverse learners effectively is not well
understood. This dissertation presents three video solutions, aiming to provide an overall view of
designing, organizing, disseminating, and assessing geoscience videos for various scales and
learning environments
Complex Permittivity of Advanced Dielectrics Across the 6G Frequency Band
No full textAs industry moves to higher frequencies with the use of 5G telecommunication (up to 38 GHz)
and automotive radar (up to 77 GHz), electrical characteristics for a wide range of materials are
needed. To potentially employ these materials with millimeter-wave ICs (MMICs), quantitative
knowledge of the materials’ dielectric properties across a broad millimeter-wave band is
necessary. To solve this, we created a nondestructive measurement of the complex relative
permittivity, εr = ε′ + jε′′, on some possible MMIC packaging materials, ranging from epoxy
composites to liquid crystal polymers as well as laminates and bond plys. Measurements using
phase-sensitive transmission over the WR3, WR5, and WR8 frequency bands (90 to 325 GHz)
show that ε′ can vary significantly in traditional packaging materials. For example, a known
material at low frequencies, Vectra A130, is shown to be dispersive at high frequencies with a
slope of 0.17%/GHz. Thus, across the WR5 band the ε′ changes from about 1.95 at 140 GHz to
about 2.8 at 220 GHz. Results like this could be used to model the performance of packaged
MMICs in HFSS and to design composites or other types of MMIC packaging material to have
tailored values of ε′ and in the loss tangent for better system results. As an example of modeling,
electromagnetic modeling results for an in-package horn antenna molded from one of the
dielectric composites measured are presented