Matching Items (12,524)
Filtering by
- Creators: Arizona State University
- Member of: Theses and Dissertations
Description
Modern digital applications have significantly increased the leakage of private and sensitive personal data. While worst-case measures of leakage such as Differential Privacy (DP) provide the strongest guarantees, when utility matters, average-case information-theoretic measures can be more relevant. However, most such information-theoretic measures do not have clear operational meanings. This dissertation addresses this challenge.
This work introduces a tunable leakage measure called maximal $\alpha$-leakage which quantifies the maximal gain of an adversary in inferring any function of a data set. The inferential capability of the adversary is modeled by a class of loss functions, namely, $\alpha$-loss. The choice of $\alpha$ determines specific adversarial actions ranging from refining a belief for $\alpha =1$ to guessing the best posterior for $\alpha = \infty$, and for the two specific values maximal $\alpha$-leakage simplifies to mutual information and maximal leakage, respectively. Maximal $\alpha$-leakage is proved to have a composition property and be robust to side information.
There is a fundamental disjoint between theoretical measures of information leakages and their applications in practice. This issue is addressed in the second part of this dissertation by proposing a data-driven framework for learning Censored and Fair Universal Representations (CFUR) of data. This framework is formulated as a constrained minimax optimization of the expected $\alpha$-loss where the constraint ensures a measure of the usefulness of the representation. The performance of the CFUR framework with $\alpha=1$ is evaluated on publicly accessible data sets; it is shown that multiple sensitive features can be effectively censored to achieve group fairness via demographic parity while ensuring accuracy for several \textit{a priori} unknown downstream tasks.
Finally, focusing on worst-case measures, novel information-theoretic tools are used to refine the existing relationship between two such measures, $(\epsilon,\delta)$-DP and R\'enyi-DP. Applying these tools to the moments accountant framework, one can track the privacy guarantee achieved by adding Gaussian noise to Stochastic Gradient Descent (SGD) algorithms. Relative to state-of-the-art, for the same privacy budget, this method allows about 100 more SGD rounds for training deep learning models.
This work introduces a tunable leakage measure called maximal $\alpha$-leakage which quantifies the maximal gain of an adversary in inferring any function of a data set. The inferential capability of the adversary is modeled by a class of loss functions, namely, $\alpha$-loss. The choice of $\alpha$ determines specific adversarial actions ranging from refining a belief for $\alpha =1$ to guessing the best posterior for $\alpha = \infty$, and for the two specific values maximal $\alpha$-leakage simplifies to mutual information and maximal leakage, respectively. Maximal $\alpha$-leakage is proved to have a composition property and be robust to side information.
There is a fundamental disjoint between theoretical measures of information leakages and their applications in practice. This issue is addressed in the second part of this dissertation by proposing a data-driven framework for learning Censored and Fair Universal Representations (CFUR) of data. This framework is formulated as a constrained minimax optimization of the expected $\alpha$-loss where the constraint ensures a measure of the usefulness of the representation. The performance of the CFUR framework with $\alpha=1$ is evaluated on publicly accessible data sets; it is shown that multiple sensitive features can be effectively censored to achieve group fairness via demographic parity while ensuring accuracy for several \textit{a priori} unknown downstream tasks.
Finally, focusing on worst-case measures, novel information-theoretic tools are used to refine the existing relationship between two such measures, $(\epsilon,\delta)$-DP and R\'enyi-DP. Applying these tools to the moments accountant framework, one can track the privacy guarantee achieved by adding Gaussian noise to Stochastic Gradient Descent (SGD) algorithms. Relative to state-of-the-art, for the same privacy budget, this method allows about 100 more SGD rounds for training deep learning models.
ContributorsLiao, Jiachun (Author) / Sankar, Lalitha (Thesis advisor) / Kosut, Oliver (Committee member) / Zhang, Junshan (Committee member) / Dasarathy, Gautam (Committee member) / Arizona State University (Publisher)
Created2020
Description
Few studies bridge workplace engagement and employee voice with internal communication. This analysis builds upon both the crucial concept of employee engagement and its implications for communication professionals and leaders. Further, it calls for more strategic integration of upward employee voice in internal communications. By examining factors that support communication (in two directions) and especially upward employee voice, researchers examine a case study of an intranet site at a major academic research institute. Factors that support employee expression are compared with data streams from both user survey and website analytics. The results point to voice-inducing techniques include projecting critical mass, fostering trust, and emphasizing intranet usefulness and rewards. By enriching workplace communications, voice can strengthen the employee’s ability to contribute, connect leaders with a source for direct feedback, and help employers be more responsive and nimbler.
ContributorsKurth, Julie (Author) / Maid, Barry (Thesis advisor) / Brumberger, Eva (Committee member) / D'Angelo, Barbara (Committee member) / Arizona State University (Publisher)
Created2020
Description
In a multi-robot system, locating a team robot is an important issue. If robots
can refer to the location of team robots based on information through passive action
recognition without explicit communication, various advantages (e.g. improving security
for military purposes) can be obtained. Specifically, when team robots follow
the same motion rule based on information about adjacent robots, associations can
be found between robot actions. If the association can be analyzed, this can be a clue
to the remote robot. Using these clues, it is possible to infer remote robots which are
outside of the sensor range.
In this paper, a multi-robot system is constructed using a combination of Thymio
II robotic platforms and Raspberry pi controllers. Robots moving in chain-formation
take action using motion rules based on information obtained through passive action
recognition. To find associations between robots, a regression model is created using
Deep Neural Network (DNN) and Long Short-Term Memory (LSTM), one of state-of-art technologies.
The input data of the regression model is divided into historical data, which
are consecutive positions of the robot, and observed data, which is information about the
observed robot. Historical data is sequence data that is analyzed through the LSTM
layer. The accuracy of the regression model designed using DNN can vary depending
on the quantity and quality of the input. In this thesis, three different input situations
are assumed for comparison. First, the amount of observed data is different, second, the
type of observed data is different, and third, the history length is different. Comparative
models are constructed for each case, and prediction accuracy is compared to analyze
the effect of input data on the regression model. This exploration validates that these
methods from deep learning can reduce the communication demands in coordinated
motion of multi-robot systems
can refer to the location of team robots based on information through passive action
recognition without explicit communication, various advantages (e.g. improving security
for military purposes) can be obtained. Specifically, when team robots follow
the same motion rule based on information about adjacent robots, associations can
be found between robot actions. If the association can be analyzed, this can be a clue
to the remote robot. Using these clues, it is possible to infer remote robots which are
outside of the sensor range.
In this paper, a multi-robot system is constructed using a combination of Thymio
II robotic platforms and Raspberry pi controllers. Robots moving in chain-formation
take action using motion rules based on information obtained through passive action
recognition. To find associations between robots, a regression model is created using
Deep Neural Network (DNN) and Long Short-Term Memory (LSTM), one of state-of-art technologies.
The input data of the regression model is divided into historical data, which
are consecutive positions of the robot, and observed data, which is information about the
observed robot. Historical data is sequence data that is analyzed through the LSTM
layer. The accuracy of the regression model designed using DNN can vary depending
on the quantity and quality of the input. In this thesis, three different input situations
are assumed for comparison. First, the amount of observed data is different, second, the
type of observed data is different, and third, the history length is different. Comparative
models are constructed for each case, and prediction accuracy is compared to analyze
the effect of input data on the regression model. This exploration validates that these
methods from deep learning can reduce the communication demands in coordinated
motion of multi-robot systems
ContributorsKang, Sehyeok (Author) / Pavlic, Theodore P (Thesis advisor) / Richa, Andréa W. (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2020
Description
One of the theoretical cores and values of good governance is the accountability of public employees, where the citizens expect the public employees to maintain professional standards, avoid conflicts of interest, respect the principles of fair and impartial treatment, and use public money wisely. However, are these unique moral standards to which only public employees are held? The dissertation seeks to examine how the public evaluates the unethical behaviors of public and private leaders differently to better understand the sources of public and private sector differences in the public’s normative evaluations.
Based on a randomized online vignette experiment with 1,569 respondents residing in the United States collected in Amazon’s Mechanical Turk platform, the dissertation confirms that public authorities face different levels of public tolerance relative to business managers. More specifically, the unethical behaviors of a public manager are less likely to be tolerated than the same misconduct of a business manager, while ethical offenses of elected officials are least likely to be tolerated by the public. However, the public is relatively much less tolerant of public managers’ and elected officials’ petty violations relative to business managers than they do for more egregious violations of public authorities.
The dissertation further finds that public evaluations are contingent upon the respondents’ work experience in different sectors. Individuals working in government are more likely to be tolerant of petty unethical behaviors, regardless of whom they evaluate, but they become much less tolerant of public managers’ and elected officials’ grand ethical violations. The longer individuals work in for-profit organizations, the less likely they are to tolerate public authorities’ petty violations of organizational rules while consistently being more accepting of the unethical behaviors of business managers.
Using an experimental design, the dissertation finds the importance of a fair and legitimate use of tax money in the public’s moral evaluations of public leadership and further discusses the potential sources of public skepticism of the public sector. Furthermore, the public and private sector comparison provides theoretical and practical implications for ethics reform in the era of collaborative governance.
Based on a randomized online vignette experiment with 1,569 respondents residing in the United States collected in Amazon’s Mechanical Turk platform, the dissertation confirms that public authorities face different levels of public tolerance relative to business managers. More specifically, the unethical behaviors of a public manager are less likely to be tolerated than the same misconduct of a business manager, while ethical offenses of elected officials are least likely to be tolerated by the public. However, the public is relatively much less tolerant of public managers’ and elected officials’ petty violations relative to business managers than they do for more egregious violations of public authorities.
The dissertation further finds that public evaluations are contingent upon the respondents’ work experience in different sectors. Individuals working in government are more likely to be tolerant of petty unethical behaviors, regardless of whom they evaluate, but they become much less tolerant of public managers’ and elected officials’ grand ethical violations. The longer individuals work in for-profit organizations, the less likely they are to tolerate public authorities’ petty violations of organizational rules while consistently being more accepting of the unethical behaviors of business managers.
Using an experimental design, the dissertation finds the importance of a fair and legitimate use of tax money in the public’s moral evaluations of public leadership and further discusses the potential sources of public skepticism of the public sector. Furthermore, the public and private sector comparison provides theoretical and practical implications for ethics reform in the era of collaborative governance.
ContributorsJung, Jiwon (Author) / Bozeman, Barry (Thesis advisor) / Bretschneider, Stuart (Committee member) / Corley, Elizabeth (Committee member) / Arizona State University (Publisher)
Created2020
Description
Recent research finds that there is significant variation in stock market participation by state and suggests that there might be state-specific factors that determine household stock market participation in the United States. Using household survey data, I examine how accounting quality of public companies at the state level affects households’ stock market participation decisions. I find that households residing in states where local public companies have better accounting quality are more likely to invest in stocks. Moreover, those households invest greater amounts of their wealth in the stock market. Cross-sectional tests find that the effect of accounting quality on stock market participation is more pronounced for less affluent and less educated households, consistent with prior findings that lacking familiarity with and trust in the stock market is an important factor deterring those types of households from stock investments. In state-level tests, I find that these household outcomes affect income inequality, which is less severe in states where high public-firm accounting quality spurs more stock market participation by poorer households. Conversely, in states where public firms have lower accounting quality, stock market participation among poorer households is less common, and a larger share of high equity returns accrues to richer households, exacerbating income inequality.
ContributorsKim, Min (Author) / Huang, Xiaochuan (Thesis advisor) / Rykaczewski, Maria (Committee member) / White, Roger (Committee member) / Arizona State University (Publisher)
Created2020
Description
What causes social systems to resist change? Studies of the emergence of social complexity in archaeology have focused primarily on drivers of change with much less emphasis on drivers of stability. Social stability, or the persistence of social systems, is an essential feature without which human society is not possible. By combining quantitative modeling (Exponential Random Graph Modeling) and the comparative archaeological record where the social system is represented by networks of relations between settlements, this research tests several hypotheses about social and geographic drivers of social stability with an explicit focus on a better understanding of contexts and processes that resist change. The Valencian Bronze Age in eastern Spain along the Mediterranean, where prior research appears to indicate little, regional social change for 700 years, serves as a case study.
The results suggest that social stability depends on a society’s ability to integrate change and promote interdependency. In part, this ability is constrained or promoted by social structure and the different, relationship dependencies among individuals that lead to a particular social structure. Four elements are important to constraining or promoting social stability—structural cohesion, transitivity and social dependency, geographic isolation, and types of exchange. Through the framework provided in this research, an archaeologist can recognize patterns in the archaeological data that reflect and promote social stability, or lead to collapse.
Results based on comparisons between the social networks of the Northern and Southern regions of the Valencian Bronze Age show that the Southern Region’s social structure was less stable through time. The Southern Region’s social structure consisted of competing cores of exchange. This type of competition often leads to power imbalances, conflict, and instability. Strong dependencies on the neighboring Argaric during the Early and Middle Bronze Ages and contributed to the Southern Region’s inability to maintain social stability after the Argaric collapsed. Furthermore, the Southern Region participated in the exchange of more complex technology—bronze. Complex technologies produce networks with hub and spoke structures highly vulnerable to collapse after the destruction of a hub. The Northern Region’s social structure remained structurally cohesive through time, promoting social stability.
The results suggest that social stability depends on a society’s ability to integrate change and promote interdependency. In part, this ability is constrained or promoted by social structure and the different, relationship dependencies among individuals that lead to a particular social structure. Four elements are important to constraining or promoting social stability—structural cohesion, transitivity and social dependency, geographic isolation, and types of exchange. Through the framework provided in this research, an archaeologist can recognize patterns in the archaeological data that reflect and promote social stability, or lead to collapse.
Results based on comparisons between the social networks of the Northern and Southern regions of the Valencian Bronze Age show that the Southern Region’s social structure was less stable through time. The Southern Region’s social structure consisted of competing cores of exchange. This type of competition often leads to power imbalances, conflict, and instability. Strong dependencies on the neighboring Argaric during the Early and Middle Bronze Ages and contributed to the Southern Region’s inability to maintain social stability after the Argaric collapsed. Furthermore, the Southern Region participated in the exchange of more complex technology—bronze. Complex technologies produce networks with hub and spoke structures highly vulnerable to collapse after the destruction of a hub. The Northern Region’s social structure remained structurally cohesive through time, promoting social stability.
ContributorsCegielski, Wendy Hope (Author) / Barton, Michael (Thesis advisor) / Kintigh, Keith (Committee member) / Coudart, Anick (Committee member) / Bernabeu-Auban, Joan (Committee member) / Arizona State University (Publisher)
Created2020
Description
In the United States, the profession of Law Enforcement is facing a workforce crisis. There are fewer applicants applying for policing jobs than there was just a decade ago. To worsen the problem, many officers are leaving the profession in less than five years. The Arizona State University Police Department is no exception to this problem. Police employees leave the department for a variety of reasons but among them is a conflict with their supervisor in the area of organizational justice. There is a gap in the training of first-line supervisors in policing as a whole as it pertains to organizational justice and how to implement it within their workgroups. Organizational Justice Theory includes the constructs of distributive justice, procedural justice, informational justice, and interpersonal justice. This mixed-methods study tested the assumption that organizational justice training with first-line supervisors at Arizona State University Police Department would have an effect on their self-efficacy and implementation of organizational justice practices and therefore improve relationships with their subordinates. Results of the study showed a single eight-hour class on Organizational Justice had no effect on the self-efficacy or implementation of organizational practices by first-line supervisors within the timeframe of the study. Like the supervisors, there was also no statistically significant effect on the employees and their belief that their supervisors were practicing organizational justice within their workgroups.
ContributorsThompson, Michael Lloyd (Author) / Judson, Eugene (Thesis advisor) / Buss, Ray R (Committee member) / Scott, Michael (Committee member) / Arizona State University (Publisher)
Created2020
Description
The number of refugees experiencing displacement is 25.9 million worldwide, with the majority in the last 7 years from Syria. While international government organizations and researchers have called for assessment of refugee health and wellness, research in this vulnerable population is limited. This dissertation is built around humanizing refugee research on health and wellness. The introduction in Chapter 1 provides an overview for the three resulting chapters which are (a) a grounded theory study to gain insight into the lives of Syrian refugees living in displacement; (b) a systematic literature review on wellness in Syrian refugees in displacement; and (c) a concept analysis to examine wellness from the perspective of Syrian refugee women within the context of displacement. Chapter 5 includes the summary, discussion, and recommendations for future research.
Chapter 2 consists of three themes which shaped the lives of Syrian refugees during displacement: (a) assets and deficits; (b) official obstacles and supports; and (c) unofficial obstacles and supports. Health emerged as a priority for the refugees which included many dimensions related to the quality of their health and health needs. The results of Chapter 2 precipitated in using wellness as a holistic lens to view Syrian refugee’s health and health needs in Chapter 3. The results of Chapter 3 added a more holistic view of Syrian refugee health, while highlighting the need for improved research methods addressing wellness in Syrian refugees. Chapter 4 clarifies and defines wellness from the perspective of Syrian refugee women.
Chapter 2 consists of three themes which shaped the lives of Syrian refugees during displacement: (a) assets and deficits; (b) official obstacles and supports; and (c) unofficial obstacles and supports. Health emerged as a priority for the refugees which included many dimensions related to the quality of their health and health needs. The results of Chapter 2 precipitated in using wellness as a holistic lens to view Syrian refugee’s health and health needs in Chapter 3. The results of Chapter 3 added a more holistic view of Syrian refugee health, while highlighting the need for improved research methods addressing wellness in Syrian refugees. Chapter 4 clarifies and defines wellness from the perspective of Syrian refugee women.
ContributorsWofford, Danielle (Author) / Komnenich, Pauline (Thesis advisor) / Fleury, Julie (Thesis advisor) / Klimek, Barbara (Committee member) / Arizona State University (Publisher)
Created2020
Description
This interpretive dissertation study sought to understand what happened when a seventh-grade teacher introduced multimodal concepts and texts into his English Language Arts classroom. Multimodal texts contain linguistic features (words and sentences) but also images and graphic design features. The classroom teacher described himself as a novice with regards to multimodal literacies instruction and had previously focused predominantly on written or spoken texts. Motivating his decision to design and enact a multimodal literacies pedagogy was his belief that students needed to garner experience interpreting and composing the kinds of texts that populated his students’ social worlds. Therefore, I asked: What happened when multimodal narratives were used as mentor texts in a seventh-grade English Language Arts classroom? Drawing from ethnographic and case study methods, I observed and gathered data regarding how the teacher and his students enacted and experienced an eight-week curriculum unit centered on multimodal concepts and multimodal texts. My findings describe the classroom teacher’s design decisions, the messiness that occurred as the classroom was (re)made into a classroom community that valued modes beyond written and spoken language, and the students’ experiences of the curriculum as classroom work, lifework, play, and drudgery. Based on my findings, I developed six assertions: (1) when designing and enacting multimodal literacies curriculum for the first time, exposing students to a wide range of multimodal texts took precedence; (2) adapted and new multimodal literacy practices began to emerge, becoming valued practices over time; (3) literacy events occurred without being grounded in literacy practices; (4) in a classroom dedicated to writing, modes of representation and communication and their associated tools and materials provided students with resources for use in their own writing/making; (5) the roles of the teacher and his students underwent change as modal expertise became sourced from across the classroom community; and (6) students experienced the multimodal literacies curriculum as play, classroom work, lifework, and drudgery. The dissertation study concludes with implications for teachers and researchers looking to converge multimodality theory with pedagogical practices and maps future research possibilities.
ContributorsReid, Stephanie Francesca (Author) / Serafini, Frank (Thesis advisor) / Moses, Lindsey (Committee member) / Marsh, Josephine (Committee member) / Williams, Wendy R. (Committee member) / Arizona State University (Publisher)
Created2020
Description随着计算机技术、互联网和云计算的高速发展，互联网＋、大数据、平台战略、长尾理论、生态圈、区块链等正在颠覆传统商业模式的运作逻辑，网络化、移动化、平台化趋势逐渐清晰。本文聚焦“互联网+”与会展平台相互融合背景下创新性数字化现代会展平台商业模式，以国内智慧会展行业领头企业——欧马腾为例，深入剖析“互联网+”赋予会展平台新的价值和成长空间，并以数据赋能为切入点，从基于大数据技术的项目监理实践、基于人工智能技术的智能营销、基于大数据的绿色生态平台建设为典型场景，系统阐述互联网会展平台成长和价值背后的重要推动作用。
研究结果发现：第一，互联网技术是欧马腾商业模式创新的重要技术保障，并为其提供了社群营销思维、大数据思维和去中心化理念，推动了欧马腾商业模式变革；第二，大数据技术是欧马腾盈利快速增长的有利支撑。这主要在于欧马腾采用大数据技术对客户售前、售中、售后进行动态跟踪，通过技术手段不断完善客户服务体系和风险控制体系，提升客户的服务体验，促使欧马腾的市场认可度逐渐上升，成为国内展览行业翘楚，品牌优势不断凸显；第三，大数据赋能欧马腾风险控制，近年来欧马腾成功的审图监理项目风险事件率为0背后的核心要素为大数据技术在审图监理项目中的应用，这充分体现了欧马腾数据赋能风险控制的成功典范；第四，人工智能赋能会展行业营销模式创新变革，欧马腾以“人工智能+”新会展生态圈为切入点，构建了智慧营销，助力其营销模式变革和商业模式转型；第五，绿色会展平台助力欧马腾价值发现创造，欧马腾的绿色平台建设能够增强现有客户再次使用的意愿，即提升欧马腾的客户黏性，从而发现和创造企业价值。
本文的研究对我国会展相关企业转型、资源整合、快速发展、可持续发展等具有重要的理论参考价值和实践借鉴。
关键词： 价值创造；数据赋能；互联网会展平台；绿色会展
研究结果发现：第一，互联网技术是欧马腾商业模式创新的重要技术保障，并为其提供了社群营销思维、大数据思维和去中心化理念，推动了欧马腾商业模式变革；第二，大数据技术是欧马腾盈利快速增长的有利支撑。这主要在于欧马腾采用大数据技术对客户售前、售中、售后进行动态跟踪，通过技术手段不断完善客户服务体系和风险控制体系，提升客户的服务体验，促使欧马腾的市场认可度逐渐上升，成为国内展览行业翘楚，品牌优势不断凸显；第三，大数据赋能欧马腾风险控制，近年来欧马腾成功的审图监理项目风险事件率为0背后的核心要素为大数据技术在审图监理项目中的应用，这充分体现了欧马腾数据赋能风险控制的成功典范；第四，人工智能赋能会展行业营销模式创新变革，欧马腾以“人工智能+”新会展生态圈为切入点，构建了智慧营销，助力其营销模式变革和商业模式转型；第五，绿色会展平台助力欧马腾价值发现创造，欧马腾的绿色平台建设能够增强现有客户再次使用的意愿，即提升欧马腾的客户黏性，从而发现和创造企业价值。
本文的研究对我国会展相关企业转型、资源整合、快速发展、可持续发展等具有重要的理论参考价值和实践借鉴。
关键词： 价值创造；数据赋能；互联网会展平台；绿色会展
ContributorsWang, Xiang (Author) / Gu, Bin (Thesis advisor) / Hu, Jie (Thesis advisor) / Zheng, Zhiqiang (Committee member) / Arizona State University (Publisher)
Created2020