12466 research outputs found
Sort by
Securing Voice Processing Systems from Malicious Audio Attacks
In an era where voice processing systems have become an integral part of our daily lives, it is of paramount importance to ensure their security and protection against malicious voice attacks. This dissertation focuses on understanding the vulnerabilities inherent in voice processing systems and presenting effective defense methods to safeguard these critical systems from exploitation. Our exploration begins by harnessing the power of the spectrum compensation technique, which serves as a cornerstone in identifying potential attack surfaces and devising appropriate defense strategies. Through meticulous analysis, we uncover a previously unknown form of voice attack - modulated replay attack. This insidious attack leverages the spectrum compensation technique to neutralize spectrum distortion caused by loudspeakers, rendering conventional spectrum-based defense methods ineffective. Our research does not stop at exposing this threat. We propose a robust defense method that combines both frequency-domain and time-domain features to effectively combat modulated replay attacks. Expanding our horizons beyond the frequency and time domains, we explore the physical properties of voice propagation. In the context of driverless car scenarios, we harness these physical properties as well as time-frequency properties to identify legitimate voice commands, enhancing the security of voice interactions in vehicular environments. In our pursuit of comprehensive defense strategies, we shed light on the versatility of spectrum compensation technique. Not only does it provide attack support for modulated replay attacks, but it also delivers defense capability in defeating spectrum reduction attacks - a practical menace to content moderation systems. The structure of this dissertation is thoughtfully designed to guide readers through the intricacy of voice processing system vulnerabilities and their corresponding defenses. In Chapter 1, we commence with an illuminating chapter that provides essential background information and a thorough review of the existing literature about attacks against voice processing systems. Building upon this foundation, Chapter 2 unveils the modulated replay attacks, showcases the spectrum compensation technique, and presents our innovative two-domain defense method. Chapter 3 represents an in-vehicle secure automatic speech recognition system. This system has leveraged multiple audio attributes, including time-domain features, frequency features, and the physical properties of voice propagation, to ensure the security of voice commands within the dynamic environment of driverless cars. Chapter 4 introduces an acoustic compensation system as a new defense against the spectrum reduction attacks that are a big threat to the content moderation systems. Through the deployment of this system, we reinforce the robustness and reliability of content filtering processes, maintaining the sanctity of online platforms. Finally, in Chapter 5, we conclude our work by summarizing the insights gained and emphasizing the importance of ongoing research. We outline potential future directions that may further strengthen voice processing systems against new emerging threats, ensuring the continued security and trustworthiness of these critical technologies. It is our fervent hope that this dissertation serves as an inspiring resource, enlightening readers about the vulnerabilities of voice processing systems and empowering them with the knowledge to develop effective defense strategies. We invite you to embark on this exploration, fostering a safer and more secure landscape for voice-based interactions
Detecting Great Power Competition Through Geospatial Analysis: A North American Arctic Case Study
This thesis has been embargoed for 5 years and will not be available until April 2026 at the earliest.The current era of Great Power Competition (GPC) between the People's Republic of China (PRC), the Russian Federation (Russia), and the United States is characterized by increased use of \hybrid threats." These are actions, short of military force, that are designed to fall under existing detection and response thresholds and compromise existing security norms and decision making processes. National security scholars and practitioners widely agree that the ability of the United States and its allies to detect and respond to these hybrid threats is limited at best, and that the Indication and Warning (I & W) intelligence function, designed to prevent strategic surprise that fundamentally alters policy, plans, and assumptions about the security environment, has atrophied. This research explores how geospatial science, through the discipline of Geospatial Intelligence (GEOINT) can detect, monitor, and provide I &Wintelligence that prevents strategic surprise from hybrid threats. Specifically, this thesis applied a novel Strategic Intelligence Framework (SIF) to standard I & W intelligence practices to identify, analyze, and visualize PRC activities that carried hybrid threat characteristics within a U.S./Canadian Arctic and circumpolar study area. Through incorporating local spatial context via the Getis-Ord Gi* statistic, as well as the strength of the hybrid threat \signal," this case study successfully identified and mapped higher and lower threat regions using kernel density estimation (KDE) in the form of a Mesoscale Operational Situational Awareness Intelligence Composite (MOSAIC). The success of this case study shows that the SIF and MOSAIC are powerful tools for detecting, analyzing, and warning about the collective impact of hybrid threats.2026-04-3
Dispelling the Shadows of Shame and Victimhood: Susan P. Conner on the Complexities of Marginalized Women in Eighteenth-Century Revolutionary France
It is a privilege and a joy to write on behalf of Susan Conner’s fine work as an historian. I am grateful to contribute to this session that considers the impact of Susan in her field. I am a lucky colleague and friend particularly for the deep and lasting impact of Susan’s scholarship and guidance. Her role of historian transcended the classroom in the multiple institutions where she taught and was revered as a beloved professor. Over the years Conner served as Associate Dean of the College of Humanities, Social & Behavioral Science at Central Michigan University, Vice-President for Academic Affairs at Florida Southern College, and ultimately Provost and Professor of History at Albion College. She received numerous awards and accolades throughout her career including a Civilian Service Medal from the U.S. Department of the Army as well as the Legion of Merit Award from the International Napoleonic Society. Conner served as a lifelong member of the American Association of University of Women and on a variety of boards including the Napoleonic Society and the Consortium on the Revolutionary Era. Her influence, both pedagogical, administrative, and collegial was broad and deep and thoughtful. One of the first lessons I learned as a graduate History student was to avoi
Discovering Blood Drinker Kerblay and a French Émigré Network in South Carolina
Frenchmen and women of all classes, aristocrats to artisans, fled political upheaval and revolutionary excesses in their nation between 1789 and 1815. Of the estimated thousands of exiles who sought safety in nearby Germany, Netherlands, and Great Britain a substantial number were aristocrats and nobles who expected to reclaim their lands soon. Others who sought more distant refuge in British North America, and the United States sought to create a new identities in the frontier regions of a distant continent. Not all aristocrats could support themselves, but pragmatic refugees of all social ranks created opportunities for themselves and others.1 Historians have studied French refugee assimilation and acculturation in New York, Maryland, Pennsylvania, and New Orleans.2 Historian Thomas Sosnowski observed that in South Carolina, most refugees were from Saint-Domingue. These refugees left traces of their new lives in legal documents and civic records. Those who left wills, inventories, and business records, and claims for indemnity, reveal a variety of backgrounds, relationships, and kinship in Charleston. A closer look at four representative French refugees illustrates their backgrounds, networks of kinship and commerce, and the varying degrees of their assimilation in frontier South Carolina. The term frontier is used loosely, because for these immigrants even the Charleston metropolis was a frontier to them
Characterizing the Aquatic Locomotor Evolution of Archaeocetes Utilizing Post-Cranial Geometric Morphometrics
Cetaceans are one of the most unusual and divergent forms of mammals in terms of both their morphology and ecology (Uhen, 2010). Numerous methods have been used to characterize the locomotion and locomotor evolution of cetaceans, from comparing fossil morphology to modern mammalian analogues (Thewissen and Fish, 1997) to observing the microstructure and osteology of archaeocete postcrania (Houssaye et al.,
2015). I used three-dimensional landmark-based geometric morphometrics to analyze and compare the shape of archaeocete innominata and to investigate the controls on their morphology. While I had hypothesized that the shape of the innominate was primarily driven by aquatic locomotory mode, my results indicated that neither locomotory mode, nor the secondary factors of phylogenetic affinity and robustness fully explained the shape variability. This suggests that more of the postcranial skeleton, such as the hindlimb and vertebral column, must be looked at in conjunction to comprehend the drivers of innominate shape change
I ❤️ IQ: A Shader Graphing Calculator for Signed Distance Functions (SDFs)
Shaders are useful in real time graphics and high performance computing applications as they specify computation to be run in parallel on the GPU. In the community, these shaders are often described in mathematical syntax and translating that math to executable code can be tedious and error prone. We present I ❤️ IQ, a graphing calculator of sorts for signed distance functions (SDFs) and materials that expedites the process of prototyping shaders. It provides a math-like syntax using I ❤️ LA and comes with a built-in raycasting architecture, removing the overhead of translation and implementation details. I ❤️ IQ is designed to be responsive and interactive. The system automatically detects free parameters and lets users tweak them using slider controls, allowing for seamless manipulation of the scene in real time. The code generated by I ❤️ IQ can then be exported to third party programs and used outside of the I IQ environment. The I ❤️ IQ repository can be found at https://github.com/HenroKriel/heartdown
Identification and Analysis of the Role of Chitinase Substrate in Francisella Novicida
This thesis has been embargoed for 2 years. It will not be available until August 2023 at the earliest.Francisella tularensis is a gram negative facultative intracellular pathogen which is a class A biothreat according to the CDC. This bacteria codes for two chitinases (ChiA and ChiB) and one chitin binding protein (CbpA). Chitin is the most abundant oligosaccharide in marine environments and in the exoskeleton of many insects. In order to digest this material, microorganisms produce chitinases capable of cleaving this polymer. The chitinases in Francisella novicida have previously been found by our lab to negatively regulate the bacterial biofilm, most likely cleaving the extracellular polysaccharide substance (EPS) with β-1,4 glycosidic linkages. These enzymes may enable the organism to use the resulting cleaved polymers as carbon and nitrogen sources for growth. Although Francisella encodes for several polysaccharide synthases, it has not been found to produce chitin. Therefore, we hypothesized that the chitinase-enzyme substrate that is self-produced in the F. novicida EPS and biofilm must be some other molecule also containing β-1,4 glycosidic linkages. The biochemical characterization of ChiB activity on various substrates will be evaluated and the role of the substrate in the biofilm will be examined. Based on the preliminary data from the van Hoek lab and other labs, we think that chitinases play important roles both in Francisella ecological persistence and macrophage mediated growth during Francisella infection. Identifying and characterizing the substrate for these chitinases will generate information on how Francisella uses them to persist in the environment and replicate in human macrophages.2023-08-2
Study of Predictive Analysis of Hospital Mortality Using ECG Signals from Heart
The hearts electrical signals play an important role in its functioning to collect deoxygenated blood and pump oxygenated blood to the rest of the body. The abnormality in the hearts muscle can cause deficiencies in electrical signal generation or passage through heart valves which can lead to the imbalance of blood flow leading to a wide range of issues from a minor abnormality to a severe outcome such as death. Therefore, it is important to measure the strength of electrical signals on patients prone to heart diseases while admitted in a hospital or remotely using wearables. In this study, I experimented with different variations of ECG signals acquired on patients in a hospital setting to study the ability of advanced deep learning methodology vs. traditional signal processing to predict mortality of a patient in the hospital. The results are based on 198 patient cohort equally split between male and female. The results from deep learning are better than the traditional methods to predict patient mortality at hospital using ECG signals of heart
Evaluation of Antithymocyte Globulin and Basiliximab as Induction Therapy in Kidney Transplant Recipients
Background
The roles of induction and maintenance therapies are becoming increasingly critical in promoting the long-term success of renal transplants and the reduction of post-surgery acute rejection. From 1998 to 2007, 78% of kidney transplant recipients (KTR) in the U.S. received immunosuppressive induction therapy [1]; fast forward to 2014, that percentage increased to 90%. [2] The three most popular antibodies used in induction therapy for KTRs in the U.S. are Antithymocyte Globulin (“Thymo”), Basiliximab (“Simulect”), and Alemtuzumab (“Campath”). According to Koyawala, et al., rabbit Thymo is used in approx. 50% of indication therapies for KTRs, Basiliximab 20%, and Campath 15%. [2] Thymo is a polyclonal antibody and lymphocytedepleting agent sourced from rabbits or horses that were immunized with human thymocytes. [3] By reducing the count of T cells, Thymo prevents KTRs’ white blood cells from rejecting the transplanted organs. Rabbit Thymo was approved in 2017 for KTRs’ induction therapy. Before then, Basiliximab was the only FDA approved induction agent for KTRs in the U.S. Basiliximab is a chimeric mouse-human antibody and an interleukin-2 receptor antagonist (IL2RA) agent that decreases patients’ T cell production by blocking certain receptors on T cells. On the other hand, Campath is a lab-produced humanized rat monoclonal antibody that works to deplete capable immune cells, T and B cells. [4]
In terms of efficacy, outcomes, and side effects after renal transplantation, Thymo and Campath are found to be most effective at preventing rejection among high-risk kidney recipients; however, Thymo is also correlated with higher rate of infections, cancer, and lymphoproliferative disease incidents. [3, 5, 6] In contrast, Basiliximab recipients showed the lowest rate of infections and side effects. Cost-wise, Basiliximab is a more cost-effective alternative to Thymo for standard and low-risk KTRs. [3, 5] An analysis of ten randomized controlled trials (over 1200 patients included) compared the efficacy of Campath relative to IL2RAs (Basiliximab and Daclizumab) and Thymo. [4] The 2012 study found that Campath resulted in a lower rate of biopsy-proven acute rejection than the induction using IL2Ras, but there was no significant difference when compared to rabbit Thymo. The study also found no apparent difference in terms of graft loss, patient death, and new-onset diabetes mellitus among the induction agents.
On the other hand, a 2019 study by Alloway et.al. examined the results of two international randomized trials (508 KTRs included) to compare the efficacy of rabbit Thymo vs. Basiliximab. [7] The study found that 1 year and 5 years after transplantation, the rate of reported treatment failures was nearly 11% lower in rabbit Thymo recipients compared to the Basiliximab recipients. Furthermore, the 10-year data of the rabbit Thymo recipients showed a 10% lower acute rejection rate than the Basiliximab recipients.
One new development in renal induction therapy is the combined use of Thymo and Basiliximab. The combined therapy has been used as a feasible alternative for patients who could not take full dosage of Thymo due to various health reasons such as thrombocytopenia, leukopenia, or cytokine release syndrome. Since Basiliximab on its own was sometimes considered too weak to prevent graft failure, an induction therapy consisting of both Thymo and Basiliximab became a plausible option. A 2023 retrospective study of 80 KTRs showed that one-year graft survival rate was not significantly different between the Basiliximab recipients and the Thymo-Basiliximab combined recipients, suggesting a positive support for the dual induction therapy consisted of low-dosage Thymo and Basiliximab. [8] On the other hand, a national study published in 2021 by Lam et. al [9] concluded that patients who received the Thymo-Basiliximab combined induction therapy have an increased risk of graft loss and mortality five years after transplantation when compared
to those who received Thymo-only induction therapy. This study analyzed data of over 150,000 KTRs from a national registry between 2005 and 2018. The other new trend worth noting is the exclusion of steroids from maintenance therapy. Due to their anti-inflammatory and immunosuppressive properties, steroids have historically been used to prevent acute rejection after renal surgeries. However, long-term steroid use could cause adverse side effects such as osteoporosis, cataract, and higher cardiovascular and infection risks. A 2016 study by Haller et al. found that the acute rejection risk was noticeably higher for patients who were subjected to steroid reduction or withdrawal. [10] However, a couple other studies found steroid reduction techniques to be safe and effective as steroid-based maintenance therapy, and that steroid withdrawal are
especially beneficial to certain patient groups, including African American recipients, sensitized patients, and pediatric patients. [11, 12]
The purpose of this study is two-fold: One is to evaluate whether the combined Thymo-Basiliximab induction therapy is as effective as other regimens (Thymo, Simulect, Campath) in preventing graft failures and patient death. The other is to examine whether the steroid-free maintenance therapy is as effective as the steroid-included maintenance therapy in graft failures and patient mortality
Measuring Attention, Working Memory and Visual Perception to Reduce the Risk of Injuries in the Construction Industry
The construction industry has consistently held one of the highest injury rates among all sectors and failure to recognize hazards due to poor selective attention, cognitive overload, and distractibility have been identified as critical human factors that lead to accidents. Considering that falls are the leading cause of deaths in the construction industry and accountable for over 33% of all construction worker deaths, this project investigated the extent to which worker characteristics (work experience, safety training, and previous injury involvement), personality dimensions (extraversion, neuroticism, conscientiousness, agreeableness, and openness to experience), working memory load and workplace conditions (e.g. time pressure) interact to influence visual attention and the identification of fall hazards. By continuously monitoring the eye movements of participants using eye-tracking technology, this study identified precursors of human error by carrying out a batch of visual search tasks to: (i) evaluate the influence of worker characteristics on visual attention and hazard identification as workers viewed 35 construction-scenario images containing 115 fall hazard areas of interests (AOIs); ii) investigate the effect of working memory load on hazard recognition for various personality traits while performing a visual attention task of identifying fall hazards across 231 AOIs and memorizing 3-digit and 6-digit strings of numbers (simulating low and high memory load conditions respectively) in a secondary cognitive task, and iii) examine the impact of time pressure on attention to fall and hand injury hazards as participants installed 27 pieces of 40 ft2 shingles standing on a low-sloped roof model 4ft wide, 6ft long and 3ft high in two experimental conditions—a baseline study without time pressure and a second manipulation with a 7-minute time limit. Multilevel analyses of data revealed that work experience, safety training and individual differences in the conscientiousness, agreeableness and openness to experience personality dimensions demonstrated significant direct associations with visual attention and superior hazard identification performance. Furthermore, residential roofers may be at a heightened risk of slip, trip, fall and hand injury hazards as a result of impaired visual attention due to time pressure. Findings have wide implications for improving safety performance and would assist organizations to assign workers to suitable tasks based on a combination of their cognitive abilities and personality variables to reduce the risk of injury among vulnerable workers whose attention may become impaired when handling multiple tasks in dynamic environments. In addition, this research is a proof of concept to construction managers on the need to prevent tight work schedules that induce time pressure and promote risk-taking, impact hazard awareness and increase workers’ susceptibility to fall and hand injury hazards