ASU Electronic Theses and Dissertations
This collection includes most of the ASU Theses and Dissertations from 2011 to present. ASU Theses and Dissertations are available in downloadable PDF format; however, a small percentage of items are under embargo. Information about the dissertations/theses includes degree information, committee members, an abstract, supporting data or media.
In addition to the electronic theses found in the ASU Digital Repository, ASU Theses and Dissertations can be found in the ASU Library Catalog.
Dissertations and Theses granted by Arizona State University are archived and made available through a joint effort of the ASU Graduate College and the ASU Libraries. For more information or questions about this collection contact or visit the Digital Repository ETD Library Guide or contact the ASU Graduate College at gradformat@asu.edu.
Displaying 1 - 10 of 425
Filtering by
- All Subjects: Machine Learning
- All Subjects: chemical engineering
Description
Sparsity has become an important modeling tool in areas such as genetics, signal and audio processing, medical image processing, etc. Via the penalization of l-1 norm based regularization, the structured sparse learning algorithms can produce highly accurate models while imposing various predefined structures on the data, such as feature groups or graphs. In this thesis, I first propose to solve a sparse learning model with a general group structure, where the predefined groups may overlap with each other. Then, I present three real world applications which can benefit from the group structured sparse learning technique. In the first application, I study the Alzheimer's Disease diagnosis problem using multi-modality neuroimaging data. In this dataset, not every subject has all data sources available, exhibiting an unique and challenging block-wise missing pattern. In the second application, I study the automatic annotation and retrieval of fruit-fly gene expression pattern images. Combined with the spatial information, sparse learning techniques can be used to construct effective representation of the expression images. In the third application, I present a new computational approach to annotate developmental stage for Drosophila embryos in the gene expression images. In addition, it provides a stage score that enables one to more finely annotate each embryo so that they are divided into early and late periods of development within standard stage demarcations. Stage scores help us to illuminate global gene activities and changes much better, and more refined stage annotations improve our ability to better interpret results when expression pattern matches are discovered between genes.
ContributorsYuan, Lei (Author) / Ye, Jieping (Thesis advisor) / Wang, Yalin (Committee member) / Xue, Guoliang (Committee member) / Kumar, Sudhir (Committee member) / Arizona State University (Publisher)
Created2013
Description
This work presents two complementary studies that propose heuristic methods to capture characteristics of data using the ensemble learning method of random forest. The first study is motivated by the problem in education of determining teacher effectiveness in student achievement. Value-added models (VAMs), constructed as linear mixed models, use students’ test scores as outcome variables and teachers’ contributions as random effects to ascribe changes in student performance to the teachers who have taught them. The VAMs teacher score is the empirical best linear unbiased predictor (EBLUP). This approach is limited by the adequacy of the assumed model specification with respect to the unknown underlying model. In that regard, this study proposes alternative ways to rank teacher effects that are not dependent on a given model by introducing two variable importance measures (VIMs), the node-proportion and the covariate-proportion. These VIMs are novel because they take into account the final configuration of the terminal nodes in the constitutive trees in a random forest. In a simulation study, under a variety of conditions, true rankings of teacher effects are compared with estimated rankings obtained using three sources: the newly proposed VIMs, existing VIMs, and EBLUPs from the assumed linear model specification. The newly proposed VIMs outperform all others in various scenarios where the model was misspecified. The second study develops two novel interaction measures. These measures could be used within but are not restricted to the VAM framework. The distribution-based measure is constructed to identify interactions in a general setting where a model specification is not assumed in advance. In turn, the mean-based measure is built to estimate interactions when the model specification is assumed to be linear. Both measures are unique in their construction; they take into account not only the outcome values, but also the internal structure of the trees in a random forest. In a separate simulation study, under a variety of conditions, the proposed measures are found to identify and estimate second-order interactions.
ContributorsValdivia, Arturo (Author) / Eubank, Randall (Thesis advisor) / Young, Dennis (Committee member) / Reiser, Mark R. (Committee member) / Kao, Ming-Hung (Committee member) / Broatch, Jennifer (Committee member) / Arizona State University (Publisher)
Created2013
Description
Cancer is the second leading cause of death in the United States and novel methods of treating advanced malignancies are of high importance. Of these deaths, prostate cancer and breast cancer are the second most fatal carcinomas in men and women respectively, while pancreatic cancer is the fourth most fatal in both men and women. Developing new drugs for the treatment of cancer is both a slow and expensive process. It is estimated that it takes an average of 15 years and an expense of $800 million to bring a single new drug to the market. However, it is also estimated that nearly 40% of that cost could be avoided by finding alternative uses for drugs that have already been approved by the Food and Drug Administration (FDA). The research presented in this document describes the testing, identification, and mechanistic evaluation of novel methods for treating many human carcinomas using drugs previously approved by the FDA. A tissue culture plate-based screening of FDA approved drugs will identify compounds that can be used in combination with the protein TRAIL to induce apoptosis selectively in cancer cells. Identified leads will next be optimized using high-throughput microfluidic devices to determine the most effective treatment conditions. Finally, a rigorous mechanistic analysis will be conducted to understand how the FDA-approved drug mitoxantrone, sensitizes cancer cells to TRAIL-mediated apoptosis.
ContributorsTaylor, David (Author) / Rege, Kaushal (Thesis advisor) / Jayaraman, Arul (Committee member) / Nielsen, David (Committee member) / Kodibagkar, Vikram (Committee member) / Dai, Lenore (Committee member) / Arizona State University (Publisher)
Created2013
Description
Liquid-liquid interfaces serve as ideal 2-D templates on which solid particles can self-assemble into various structures. These self-assembly processes are important in fabrication of micron-sized devices and emulsion formulation. At oil/water interfaces, these structures can range from close-packed aggregates to ordered lattices. By incorporating an ionic liquid (IL) at the interface, new self-assembly phenomena emerge. ILs are ionic compounds that are liquid at room temperature (essentially molten salts at ambient conditions) that have remarkable properties such as negligible volatility and high chemical stability and can be optimized for nearly any application. The nature of IL-fluid interfaces has not yet been studied in depth. Consequently, the corresponding self-assembly phenomena have not yet been explored. We demonstrate how the unique molecular nature of ILs allows for new self-assembly phenomena to take place at their interfaces. These phenomena include droplet bridging (the self-assembly of both particles and emulsion droplets), spontaneous particle transport through the liquid-liquid interface, and various gelation behaviors. In droplet bridging, self-assembled monolayers of particles effectively "glue" emulsion droplets to one another, allowing the droplets to self-assembly into large networks. With particle transport, it is experimentally demonstrated the ILs overcome the strong adhesive nature of the liquid-liquid interface and extract solid particles from the bulk phase without the aid of external forces. These phenomena are quantified and corresponding mechanisms are proposed. The experimental investigations are supported by molecular dynamics (MD) simulations, which allow for a molecular view of the self-assembly process. In particular, we show that particle self-assembly depends primarily on the surface chemistry of the particles and the non-IL fluid at the interface. Free energy calculations show that the attractive forces between nanoparticles and the liquid-liquid interface are unusually long-ranged, due to capillary waves. Furthermore, IL cations can exhibit molecular ordering at the IL-oil interface, resulting in a slight residual charge at this interface. We also explore the transient IL-IL interface, revealing molecular interactions responsible for the unusually slow mixing dynamics between two ILs. This dissertation, therefore, contributes to both experimental and theoretical understanding of particle self-assembly at IL based interfaces.
ContributorsFrost, Denzil (Author) / Dai, Lenore L (Thesis advisor) / Torres, César I (Committee member) / Nielsen, David R (Committee member) / Squires, Kyle D (Committee member) / Rege, Kaushal (Committee member) / Arizona State University (Publisher)
Created2013
Description
To further the efforts producing energy from more renewable sources, microbial electrochemical cells (MXCs) can utilize anode respiring bacteria (ARB) to couple the oxidation of an organic substrate to the delivery of electrons to the anode. Although ARB such as Geobacter and Shewanella have been well-studied in terms of their microbiology and electrochemistry, much is still unknown about the mechanism of electron transfer to the anode. To this end, this thesis seeks to elucidate the complexities of electron transfer existing in Geobacter sulfurreducens biofilms by employing Electrochemical Impedance Spectroscopy (EIS) as the tool of choice. Experiments measuring EIS resistances as a function of growth were used to uncover the potential gradients that emerge in biofilms as they grow and become thicker. While a better understanding of this model ARB is sought, electrochemical characterization of a halophile, Geoalkalibacter subterraneus (Glk. subterraneus), revealed that this organism can function as an ARB and produce seemingly high current densities while consuming different organic substrates, including acetate, butyrate, and glycerol. The importance of identifying and studying novel ARB for broader MXC applications was stressed in this thesis as a potential avenue for tackling some of human energy problems.
ContributorsAjulo, Oluyomi (Author) / Torres, Cesar (Thesis advisor) / Nielsen, David (Committee member) / Krajmalnik-Brown, Rosa (Committee member) / Popat, Sudeep (Committee member) / Arizona State University (Publisher)
Created2013
Description
Image understanding has been playing an increasingly crucial role in vision applications. Sparse models form an important component in image understanding, since the statistics of natural images reveal the presence of sparse structure. Sparse methods lead to parsimonious models, in addition to being efficient for large scale learning. In sparse modeling, data is represented as a sparse linear combination of atoms from a "dictionary" matrix. This dissertation focuses on understanding different aspects of sparse learning, thereby enhancing the use of sparse methods by incorporating tools from machine learning. With the growing need to adapt models for large scale data, it is important to design dictionaries that can model the entire data space and not just the samples considered. By exploiting the relation of dictionary learning to 1-D subspace clustering, a multilevel dictionary learning algorithm is developed, and it is shown to outperform conventional sparse models in compressed recovery, and image denoising. Theoretical aspects of learning such as algorithmic stability and generalization are considered, and ensemble learning is incorporated for effective large scale learning. In addition to building strategies for efficiently implementing 1-D subspace clustering, a discriminative clustering approach is designed to estimate the unknown mixing process in blind source separation. By exploiting the non-linear relation between the image descriptors, and allowing the use of multiple features, sparse methods can be made more effective in recognition problems. The idea of multiple kernel sparse representations is developed, and algorithms for learning dictionaries in the feature space are presented. Using object recognition experiments on standard datasets it is shown that the proposed approaches outperform other sparse coding-based recognition frameworks. Furthermore, a segmentation technique based on multiple kernel sparse representations is developed, and successfully applied for automated brain tumor identification. Using sparse codes to define the relation between data samples can lead to a more robust graph embedding for unsupervised clustering. By performing discriminative embedding using sparse coding-based graphs, an algorithm for measuring the glomerular number in kidney MRI images is developed. Finally, approaches to build dictionaries for local sparse coding of image descriptors are presented, and applied to object recognition and image retrieval.
ContributorsJayaraman Thiagarajan, Jayaraman (Author) / Spanias, Andreas (Thesis advisor) / Frakes, David (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Description
The use of petroleum for liquid-transportation fuels has strained the environment and caused the global crude oil reserves to diminish. Therefore, there exists a need to replace petroleum as the primary fuel derivative. Butanol is a four-carbon alcohol that can be used to effectively replace gasoline without changing the current automotive infrastructure. Additionally, butanol offers the same environmentally friendly effects as ethanol, but possess a 23% higher energy density. Clostridium acetobutylicum is an anaerobic bacterium that can ferment renewable biomass-derived sugars into butanol. However, this fermentation becomes limited by relatively low butanol concentrations (1.3% w/v), making this process uneconomical. To economically produce butanol, the in-situ product removal (ISPR) strategy is employed to the butanol fermentation. ISPR entails the removal of butanol as it is produced, effectively avoiding the toxicity limit and allowing for increased overall butanol production. This thesis explores the application of ISPR through integration of expanded-bed adsorption (EBA) with the C. acetobutylicum butanol fermentations. The goal is to enhance volumetric productivity and to develop a semi-continuous biofuel production process. The hydrophobic polymer resin adsorbent Dowex Optipore L-493 was characterized in cell-free studies to determine the impact of adsorbent mass and circulation rate on butanol loading capacity and removal rate. Additionally, the EBA column was optimized to use a superficial velocity of 9.5 cm/min and a resin fraction of 50 g/L. When EBA was applied to a fed-batch butanol fermentation performed under optimal operating conditions, a total of 25.5 g butanol was produced in 120 h, corresponding to an average yield on glucose of 18.6%. At this level, integration of EBA for in situ butanol recovered enabled the production of 33% more butanol than the control fermentation. These results are very promising for the production of butanol as a biofuel. Future work will entail the optimization of the fed-batch process for higher glucose utilization and development of a reliable butanol recovery system from the resin.
ContributorsWiehn, Michael (Author) / Nielsen, David (Thesis advisor) / Lin, Jerry (Committee member) / Lind, Mary Laura (Committee member) / Arizona State University (Publisher)
Created2013
Description
Alzheimer's disease (AD) is the leading neurodegenerative disease, affecting roughly 8% of people 65 years of age or older. There exists an imperative need to develop a non-invasive test for the earlier detection of AD. The use of biomarkers is a promising option that examines the toxic mechanisms and metabolic pathways that cause Alzheimer's disease, eventually leading to an early diagnostic method. This thesis presents the use of oligomeric beta-amyloid as a biomarker to detect Alzheimer's disease via a specialized enzyme-linked protein assay. Specifically, this paper details the optimization and development of a novel phage capture enzyme-linked immunosorbent assay (ELISA) that can detect the relative quantity of beta-amyloid oligomers in samples from a mouse model of AD. The objective of this thesis was to optimize a phage capture ELISA using the A4 single-chain variable fragment (scFv) to quantify the amount of beta-amyloid oligomers in various mice samples. A4 selectively recognizes a toxic oligomeric form of beta-amyloid. The level of A4-reactive oligomeric beta-amyloid was measured in triplicate in homogenized mouse brain tissue samples from eight transgenic (TG) and eight nontransgenic (NTG) animals aged five, nine, and thirteen months. There was a significant difference (p < 0.0005) between the five month TG and NTG mice. A decrease in beta-amyloid levels with the aging of the TG mice suggested that the beta-amyloid oligomers may be aggregating to form beta-amyloid fibrils. Conversely, the quantity of beta-amyloid increased with the aging of the NTG mice. This indicated that beta-amyloid oligomers may develop with normal aging.
ContributorsBrownlee, Taylor (Author) / Sierks, Michael (Thesis advisor) / Williams, Stephanie (Committee member) / Xin, Wei (Committee member) / Arizona State University (Publisher)
Created2013
Description
In today's world where sustainability is of prime importance, energy efficient method for sea water desalination and waste water treatment is gaining attention. State of art Reverse Osmosis (RO) method has high power consumption; hence people are diverting their attention towards Forward Osmosis (FO). It has been determined that the support membrane hydrophilicity plays an important role impacting the water flux through membranes in forward osmosis processes. The support layer of commercially available thin film composite RO membranes has been modified with a hydrophilic polymer Polyvinyl Alcohol (PVA). Previous research has demonstrated that PVA coating of the top selective layer of RO membranes has decreased selective layer roughness and increased selective layer hydrophilicity. The role of PVA with 2 different PVA cross-linkers: Maleic Acid (MA) and Glutaraldehyde (GA) at 2 different concentrations of 10% and 50% have been investigated. The hydrophilicity, water flux, salt flux and rejection of the neat and modified membranes in Reverse Osmosis and Forward Osmosis are measured. Maleic Acid when used with PVA at a lower degree of cross linking (10%) shows significant improvement in water flux in SW membranes in comparison to Glutaraldehyde cross-linked PVA coated membranes. This improvement is not so significantly observed in BW membranes due to its lower porosity. PVA when used with a small amount of cross-linker shows promising results in increasing the hydrophilicity of TFC membranes enabling RO membranes to be used efficiently in FO processes.
ContributorsSaraf, Aditi (Author) / Lind, Dr. Mary (Thesis advisor) / Dai, Dr. Lenore (Committee member) / Nielsen, Dr. David (Committee member) / Arizona State University (Publisher)
Created2012
Description
With the increase in computing power and availability of data, there has never been a greater need to understand data and make decisions from it. Traditional statistical techniques may not be adequate to handle the size of today's data or the complexities of the information hidden within the data. Thus knowledge discovery by machine learning techniques is necessary if we want to better understand information from data. In this dissertation, we explore the topics of asymmetric loss and asymmetric data in machine learning and propose new algorithms as solutions to some of the problems in these topics. We also studied variable selection of matched data sets and proposed a solution when there is non-linearity in the matched data. The research is divided into three parts. The first part addresses the problem of asymmetric loss. A proposed asymmetric support vector machine (aSVM) is used to predict specific classes with high accuracy. aSVM was shown to produce higher precision than a regular SVM. The second part addresses asymmetric data sets where variables are only predictive for a subset of the predictor classes. Asymmetric Random Forest (ARF) was proposed to detect these kinds of variables. The third part explores variable selection for matched data sets. Matched Random Forest (MRF) was proposed to find variables that are able to distinguish case and control without the restrictions that exists in linear models. MRF detects variables that are able to distinguish case and control even in the presence of interaction and qualitative variables.
ContributorsKoh, Derek (Author) / Runger, George C. (Thesis advisor) / Wu, Tong (Committee member) / Pan, Rong (Committee member) / Cesta, John (Committee member) / Arizona State University (Publisher)
Created2013