Matching Items (43,913)
Description
This dissertation transforms a set of system complexity reduction problems to feature selection problems. Three systems are considered: classification based on association rules, network structure learning, and time series classification. Furthermore, two variable importance measures are proposed to reduce the feature selection bias in tree models. Associative classifiers can achieve high accuracy, but the combination of many rules is difficult to interpret. Rule condition subset selection (RCSS) methods for associative classification are considered. RCSS aims to prune the rule conditions into a subset via feature selection. The subset then can be summarized into rule-based classifiers. Experiments show that classifiers after RCSS can substantially improve the classification interpretability without loss of accuracy. An ensemble feature selection method is proposed to learn Markov blankets for either discrete or continuous networks (without linear, Gaussian assumptions). The method is compared to a Bayesian local structure learning algorithm and to alternative feature selection methods in the causal structure learning problem. Feature selection is also used to enhance the interpretability of time series classification. Existing time series classification algorithms (such as nearest-neighbor with dynamic time warping measures) are accurate but difficult to interpret. This research leverages the time-ordering of the data to extract features, and generates an effective and efficient classifier referred to as a time series forest (TSF). The computational complexity of TSF is only linear in the length of time series, and interpretable features can be extracted. These features can be further reduced, and summarized for even better interpretability. Lastly, two variable importance measures are proposed to reduce the feature selection bias in tree-based ensemble models. It is well known that bias can occur when predictor attributes have different numbers of values. Two methods are proposed to solve the bias problem. One uses an out-of-bag sampling method called OOBForest, and the other, based on the new concept of a partial permutation test, is called a pForest. Experimental results show the existing methods are not always reliable for multi-valued predictors, while the proposed methods have advantages.
ContributorsDeng, Houtao (Author) / Runger, George C. (Thesis advisor) / Lohr, Sharon L (Committee member) / Pan, Rong (Committee member) / Zhang, Muhong (Committee member) / Arizona State University (Publisher)
Created2011
Description
Negative bias temperature instability (NBTI) and channel hot carrier (CHC) are important reliability issues impacting analog circuit performance and lifetime. Compact reliability models and efficient simulation methods are essential for circuit level reliability prediction. This work proposes a set of compact models of NBTI and CHC effects for analog and mixed-signal circuit, and a direct prediction method which is different from conventional simulation methods. This method is applied in circuit benchmarks and evaluated. This work helps with improving efficiency and accuracy of circuit aging prediction.
ContributorsZheng, Rui (Author) / Cao, Yu (Thesis advisor) / Yu, Hongyu (Committee member) / Bakkaloglu, Bertan (Committee member) / Arizona State University (Publisher)
Created2011
Description
ABSTRACT Early career science teachers are often assigned to classrooms with high numbers of English language learners (ELL students). As these teachers learn to become effective practitioners, the circumstances surrounding them merit a thorough examination. This study examines the longitudinal changes in Pedagogical Content Knowledge (PCK) and practices of six early career science teachers who taught in urban schools. The teachers participated in the Alternative Support for Induction Science Teachers (ASIST) program during their initial two years of teaching. Our research team followed the participants over a five-year period. This study focuses on data from Years 1, 3, and 5. The data collected included classroom observations and interviews. In addition, classroom artifacts were collected periodically for the purpose of triangulation. The analysis of the data revealed that with the support of the ASIST program, the teachers implemented inquiry lessons and utilized instructional materials that promoted academic language skills and science competencies among their ELL students. Conversely, standardized testing, teaching assignment, and school culture played a role in constraining the implementation of inquiry-based practices. The results of this study call for collaborative efforts among university science educators and school administrators to provide professional development opportunities and support for the implementation of inquiry and language practices among early career science teachers of ELL students.
ContributorsOrtega, Irasema (Author) / Luft, Julie A (Thesis advisor) / Artiles, Alfredo (Committee member) / Baker, Dale R. (Committee member) / Arizona State University (Publisher)
Created2011
Description
There is increasing interest in the medical and behavioral health communities towards developing effective strategies for the treatment of chronic diseases. Among these lie adaptive interventions, which consider adjusting treatment dosages over time based on participant response. Control engineering offers a broad-based solution framework for optimizing the effectiveness of such interventions. In this thesis, an approach is proposed to develop dynamical models and subsequently, hybrid model predictive control schemes for assigning optimal dosages of naltrexone, an opioid antagonist, as treatment for a chronic pain condition known as fibromyalgia. System identification techniques are employed to model the dynamics from the daily diary reports completed by participants of a blind naltrexone intervention trial. These self-reports include assessments of outcomes of interest (e.g., general pain symptoms, sleep quality) and additional external variables (disturbances) that affect these outcomes (e.g., stress, anxiety, and mood). Using prediction-error methods, a multi-input model describing the effect of drug, placebo and other disturbances on outcomes of interest is developed. This discrete time model is approximated by a continuous second order model with zero, which was found to be adequate to capture the dynamics of this intervention. Data from 40 participants in two clinical trials were analyzed and participants were classified as responders and non-responders based on the models obtained from system identification. The dynamical models can be used by a model predictive controller for automated dosage selection of naltrexone using feedback/feedforward control actions in the presence of external disturbances. The clinical requirement for categorical (i.e., discrete-valued) drug dosage levels creates a need for hybrid model predictive control (HMPC). The controller features a multiple degree-of-freedom formulation that enables the user to adjust the speed of setpoint tracking, measured disturbance rejection and unmeasured disturbance rejection independently in the closed loop system. The nominal and robust performance of the proposed control scheme is examined via simulation using system identification models from a representative participant in the naltrexone intervention trial. The controller evaluation described in this thesis gives credibility to the promise and applicability of control engineering principles for optimizing adaptive interventions.
ContributorsDeśapāṇḍe, Sunīla (Author) / Rivera, Daniel E. (Thesis advisor) / Si, Jennie (Committee member) / Tsakalis, Konstantinos (Committee member) / Arizona State University (Publisher)
Created2011
Description
Thousands of children are being injured every day in bicycling accidents. Interventions, like Safe Routes to School, are currently in place to combat injury rates by providing programs to teach children safe biking behaviors. In order to develop effective behavioral change programs, behavior and the components of which it is composed must be understood. Attitudes, subjective norms, and self-efficacy are important predictors of intention to perform a behavior. The purpose of this study was to ascertain the extent to which attitude, subjective norms, self-efficacy, and bike rodeo participation explain third through eighth graders' intentions to bike safely. These constructs were tested using a survey research design in a purposive sample of fifty-seven third through eighth grade students in Safe Routes to School schools in the Southwest. Students took an online survey in the computer lab at their respective schools supervised by a teacher. The study found attitudes to be comprised of three factors: happy/safe, not afraid, and calm. Overall, the model explained approximately 71% of the variance in children's intentions to bike safely, R2=.749, Adjusted R2=.713, F(7, 49)=20.854, p<.01. The significant predictors were happy/safe attitudes, subjective norms, self-efficacy, and a quadratic self-efficacy term explaining 10% (p=.019), 28% (p<.01), 18% (p<.01), and 15% (p<.01) respectively. The results of the study can be used to create future and improve current bike safety interventions for children.
ContributorsPayton, Kayla G (Author) / Rodriguez, Ariel (Thesis advisor) / Tyrrell, Timothy (Committee member) / Maruyama, Kenichi (Committee member) / Arizona State University (Publisher)
Created2011
Description
Nucleosomes are the basic repetitive unit of eukaryotic chromatin and are responsible for packing DNA inside the nucleus of the cell. They consist of a complex of eight histone proteins (two copies of four proteins H2A, H2B, H3 and H4) around which 147 base pairs of DNA are wrapped in ~1.67 superhelical turns. Although the nucleosomes are stable protein-DNA complexes, they undergo spontaneous conformational changes that occur in an asynchronous fashion. This conformational dynamics, defined by the "site-exposure" model, involves the DNA unwrapping from the protein core and exposing itself transiently before wrapping back. Physiologically, this allows regulatory proteins to bind to their target DNA sites during cellular processes like replication, DNA repair and transcription. Traditional biochemical assays have stablished the equilibrium constants for the accessibility to various sites along the length of the nucleosomal DNA, from its end to the middle of the dyad axis. Using fluorescence correlation spectroscopy (FCS), we have established the position dependent rewrapping rates for nucleosomes. We have also used Monte Carlo simulation methods to analyze the applicability of FRET fluctuation spectroscopy towards conformational dynamics, specifically motivated by nucleosome dynamics. Another important conformational change that is involved in cellular processes is the disassembly of nucleosome into its constituent particles. The exact pathway adopted by nucleosomes is still not clear. We used dual color fluorescence correlation spectroscopy to study the intermediates during nucleosome disassembly induced by changing ionic strength. Studying the nature of nucleosome conformational change and the kinetics is very important in understanding gene expression. The results from this thesis give a quantitative description to the basic unit of the chromatin.
ContributorsGurunathan, Kaushik (Author) / Levitus, Marcia (Thesis advisor) / Lindsay, Stuart (Committee member) / Woodbury, Neal (Committee member) / Yan, Hao (Committee member) / Arizona State University (Publisher)
Created2011
Description
This dissertation investigates socio-economic strategies adopted by a small craftworking community situated on the edge of one of the earliest, largest and most complex cities in Mesoamerica. The focus of investigation is San Jose 520, a hamlet located on the southeastern margin of Teotihuacan and occupied primarily during the Tlamimilolpa and Xolalpan phases (ca. A.D. 200-500). Its inhabitants were potters of low socio-economic status living in small, architecturally simple residential structures. The investigation complements much more numerous studies of higher-status groups residing in Teotihuacan's famous apartment compounds, much larger and architecturally more formal structures clustered primarily within built-up parts of the city. The founding residents of San Jose 520 might have initially been immigrants, arriving at Teotihuacan after most of the city was already filled in and occupied, and therefore settling in a spatially marginal area with limited potential for farming. Archaeological field and lab investigations demonstrate that they adopted ceramic production as a strategy of economic survival in a competitive urban system. They specialized in the manufacture of the outcurving bowl--a vessel widely used at Teotihuacan for food service and certain ritual activities. At smaller scales of production, these potters also made other types of serving and ritual vessels and figurines. Evidence relating to mortuary and domestic rituals indicates participation in a number of the rituals typical of other sectors of Teotihuacan society, but not all. The most general goal of this investigation is to improve understanding of how socially and spatially marginal peoples possessing low economic status developed and exploited viable economic niches in pre-industrial urban systems. The San Jose 520 potters appear dynamic in their economic adjustment--in part by enhancing their production system over time through the adoption of various specialized pot-making tools (some as yet undocumented for Teotihuacan), and to some extent by modifying their product line, they survived for many generations. Nevertheless, they never succeeded in significantly raising their economic status; at the time of their apparent disappearance sometime in the Xolalpan phase, these potters and their households continued to constitute a case study of urban poverty in a massive pre-industrial city.
ContributorsCabrera Cortes, Mercedes Oralia (Author) / Stark, Barbara L. (Thesis advisor) / Cowgill, George L. (Thesis advisor) / Falconer, Steven E. (Committee member) / Spielmann, Katherine A. (Committee member) / Arizona State University (Publisher)
Created2011
Description
This dissertation is focused on building scalable Attribute Based Security Systems (ABSS), including efficient and privacy-preserving attribute based encryption schemes and applications to group communications and cloud computing. First of all, a Constant Ciphertext Policy Attribute Based Encryption (CCP-ABE) is proposed. Existing Attribute Based Encryption (ABE) schemes usually incur large, linearly increasing ciphertext. The proposed CCP-ABE dramatically reduces the ciphertext to small, constant size. This is the first existing ABE scheme that achieves constant ciphertext size. Also, the proposed CCP-ABE scheme is fully collusion-resistant such that users can not combine their attributes to elevate their decryption capacity. Next step, efficient ABE schemes are applied to construct optimal group communication schemes and broadcast encryption schemes. An attribute based Optimal Group Key (OGK) management scheme that attains communication-storage optimality without collusion vulnerability is presented. Then, a novel broadcast encryption model: Attribute Based Broadcast Encryption (ABBE) is introduced, which exploits the many-to-many nature of attributes to dramatically reduce the storage complexity from linear to logarithm and enable expressive attribute based access policies. The privacy issues are also considered and addressed in ABSS. Firstly, a hidden policy based ABE schemes is proposed to protect receivers' privacy by hiding the access policy. Secondly,a new concept: Gradual Identity Exposure (GIE) is introduced to address the restrictions of hidden policy based ABE schemes. GIE's approach is to reveal the receivers' information gradually by allowing ciphertext recipients to decrypt the message using their possessed attributes one-by-one. If the receiver does not possess one attribute in this procedure, the rest of attributes are still hidden. Compared to hidden-policy based solutions, GIE provides significant performance improvement in terms of reducing both computation and communication overhead. Last but not least, ABSS are incorporated into the mobile cloud computing scenarios. In the proposed secure mobile cloud data management framework, the light weight mobile devices can securely outsource expensive ABE operations and data storage to untrusted cloud service providers. The reported scheme includes two components: (1) a Cloud-Assisted Attribute-Based Encryption/Decryption (CA-ABE) scheme and (2) An Attribute-Based Data Storage (ABDS) scheme that achieves information theoretical optimality.
ContributorsZhou, Zhibin (Author) / Huang, Dijiang (Thesis advisor) / Yau, Sik-Sang (Committee member) / Ahn, Gail-Joon (Committee member) / Reisslein, Martin (Committee member) / Arizona State University (Publisher)
Created2011
Description
In the rural, modern American West, two Manichean perspectives of the human-nature relationship have contributed to vehement environmental conflicts. Adopting developer Calvin Black and writer Edward Abbey as archetypes, I explore the endurance of these two ideologies in the redrock canyon country of southern Utah and northern Arizona. Black represents the historically dominant anthropocentric view among Euro Americans that nature ought to be domesticated and commoditized; the competing view, represented by Abbey, is eco-centric and considers the intrinsic value of the broader ecological community beyond its utilitarian function. I argue that environmental conflict in the canyon country has been driven by ideologues who espouse one of these two deeply entrenched and seemingly irreconcilable perspectives. Modern-day conflicts over wilderness, land use, and rural development are endemic, rooted in heritage and culture and driven by particular Anglo-American religious and secular beliefs that reflect differing ways of “seeing” the land. In particular these contending perspectives are reflected in the “built” landscape. Using one especially ubiquitous human imprint on the land as both trope and subject, I explore the political and cultural meanings of roads as symbols variously of progress and of exploitation. Questions of road development and public lands access became the center point of environmental conflict driven by dichotomous worldviews that demonized the opposition and its position. What developed in the last half century is a discourse dictated by categories created by ideologues. This dissertation not only explores the particular circumstances that made these environmental contests volatile in an American desert, but it also meditates broadly on the nature of environmental compromise and conflict, the place of people in "wild" landscapes, and the discontents of rural communities upended by new economic realities. This study illustrates generally how people perceive the land, the technology they wield to manipulate it, and the broader cultural and political transformations that result.
ContributorsRogers, Jedediah S. (Jedediah Smart), 1978- (Author) / Hirt, Paul W. (Thesis advisor) / Pyne, Stephen J. (Committee member) / Fixico, Donald L. (Committee member) / Arizona State University (Publisher)
Created2011
Description
Thermoelectric devices (TED's) continue to be an area of high interest in both thermal management and energy harvesting applications. Due to their compact size, reliable performance, and their ability to accomplish sub-ambient cooling, much effort is being focused on optimized methods for characterization and integration of TED's for future applications. Predictive modeling methods can only achieve accurate results with robust input physical parameters, therefore TED characterization methods are critical for future development of the field. Often times, physical properties of TED sub-components are very well known, however the "effective" properties of a TED module can be difficult to measure with certainty. The module-level properties must be included in predictive modeling, since these include electrical and thermal contact resistances which are difficult to analytically derive. A unique characterization method is proposed, which offers the ability to directly measure all device-level physical parameters required for accurate modeling. Among many other unique features, the metrology allows the capability to perform an independent validation of empirical parameters by measuring parasitic heat losses. As support for the accuracy of the measured parameters, the metrology output from an off-the-shelf TED is used in a system-level thermal model to predict and validate observed metrology temperatures. Finally, as an extension to the benefits of this metrology, it is shown that resulting data can be used to empirically validate a device-level dimensionless relationship. The output provides a powerful performance prediction tool, since all physical behavior in a performance domain is captured using a single analytical relationship and can be plotted on a singe graph.
ContributorsLofgreen, Kelly (Author) / Phelan, Patrick E (Thesis advisor) / Posner, Jonathan (Committee member) / Devasenathipathy, Shankar (Committee member) / Arizona State University (Publisher)
Created2011