Theses and Dissertations
This collection includes both ASU Theses and Dissertations, submitted by graduate students, and the Barrett, Honors College theses submitted by undergraduate students.
Displaying 21 - 30 of 96
Description
Volatile Organic Compounds (VOCs) are central to atmospheric chemistry and have significant impacts on the environment. The reaction of oxygenated VOCs with OH radicals was first studied to understand the fate of oxygenated VOCs. The rate constants of the gas-phase reaction of OH radicals with trans-2-hexenal, trans-2-octenal, and trans-2 nonenal were determined using the relative rate technique. Then the interactions between VOCs and ionic liquid surfaces were studied. The goal was to find a material to selectively detect alcohol compounds. Computational chemistry calculations were performed to investigate the interactions of ionic liquids with different classes of VOCs. The thermodynamic data suggest that 1-butyl-3-methylimindazolium chloride (C4mimCl) preferentially interacts with alcohols as compared to other classes of VOCs. Fourier transform infrared spectroscopy was used to probe the ionic liquid surface before and after exposure to the VOCs that were tested. New spectral features were detected after exposure of C4mimCl to various alcohols and a VOC mixture with an alcohol in it. The new features are characteristic of the alcohols tested. No new IR features were detected after exposure of the C4mimCl to the aldehyde, ketone, alkane, alkene, alkyne or aromatic compounds. The experimental results demonstrated that C4mimCl is selective to alcohols, even in complex mixtures. The kinetic study of the association and dissociation of alcohols with C4minCl surfaces was performed. The findings in this work provide information for future gas-phase alcohol sensor design. CO2 is a major contributor to global warming. An ionic liquid functionalized reduced graphite oxide (IL-RGO)/ TiO2 nanocomposite was synthesized and used to reduce CO2 to a hydrocarbon in the presence of H2O vapor. The SEM image revealed that IL-RGO/TiO2 contained separated reduced graphite oxide flakes with TiO2 nanoparticles. Diffuse Reflectance Infrared Fourier Transform Spectroscopy was used to study the conversion of CO2 and H2O vapor over the IL-RGO/TiO2 catalyst. Under UV-Vis irradiation, CH4 was found to form after just 40 seconds of irradiation. The concentration of CH4 continuously increased under longer irradiation time. This research is particularly important since it seems to suggest the direct, selective formation of CH4 as opposed to CO.
ContributorsGao, Tingting (Author) / Andino, Jean M (Thesis advisor) / Forzani, Erica (Committee member) / Kavazanjian, Edward (Committee member) / Arizona State University (Publisher)
Created2012
Description
The influence of temperature on soil engineering properties is a major concern in the design of engineering systems such as radioactive waste disposal barriers, ground source heat pump systems and pavement structures. In particular, moisture redistribution under pavement systems might lead to changes in unbound material stiffness that will affect pavement performance. Accurate measurement of thermal effects on unsaturated soil hydraulic properties may lead to reduction in design and construction costs. This thesis presents preliminary results of an experimental study aimed at determining the effect of temperature on the soil water characteristic curve (SWCC) and the unsaturated hydraulic conductivity function (kunsat). Pressure plate devices with volume change control were used to determine the SWCC and the instantaneous profile method was used to obtain the kunsat function. These properties were measured on two fine-grained materials subjected to controlled temperatures of 5oC, 25oC and 40oC. The results were used to perform a sensitivity analysis of the effect of temperature changes on the prediction of moisture movement under a covered area. In addition, two more simulations were performed where changes in hydraulic properties were done in a stepwise fashion. The findings were compared to field measured water content data obtained on the subgrade material of the FAA William Hughes test facility located in Atlantic City. Results indicated that temperature affects the unsaturated hydraulic properties of the two soils used in the study. For the DuPont soil, a soil with high plasticity, it was found that the water retention was higher at low temperatures for suction levels lower than about 10,000 kPa; while the kunsat functions at the three temperatures were not significantly different. For the County soil, a material with medium plasticity, it was found that it holds around 10% more degree of saturation at 5°C than that at 40°C for suction levels higher than about 1,000 kPa; while the hydraulic conductivity at 40°C was at least one order of magnitude higher than that at 5°C, for suction levels higher than 1,000 kPa. These properties were used to perform two types of numerical analyses: a sensitivity analysis and stepwise analysis. Absolute differences between predicted and field measured data were considered to be acceptable, ranging from 4.5% to 9% for all simulations. Overall results show an improvement in predictions when non-isothermal conditions were used over the predictions obtained with isothermal conditions.
ContributorsLu, Yutong (Author) / Zapata, Claudia E (Thesis advisor) / Kavazanjian, Edward (Committee member) / Houston, Sandra L. (Committee member) / Arizona State University (Publisher)
Created2015
Description
Through decades of clinical progress, cochlear implants have brought the world of speech and language to thousands of profoundly deaf patients. However, the technology has many possible areas for improvement, including providing information of non-linguistic cues, also called indexical properties of speech. The field of sensory substitution, providing information relating one sense to another, offers a potential avenue to further assist those with cochlear implants, in addition to the promise they hold for those without existing aids. A user study with a vibrotactile device is evaluated to exhibit the effectiveness of this approach in an auditory gender discrimination task. Additionally, preliminary computational work is included that demonstrates advantages and limitations encountered when expanding the complexity of future implementations.
ContributorsButts, Austin McRae (Author) / Helms Tillery, Stephen (Thesis advisor) / Berisha, Visar (Committee member) / Buneo, Christopher (Committee member) / McDaniel, Troy (Committee member) / Arizona State University (Publisher)
Created2015
Description
Our ability to understand networks is important to many applications, from the analysis and modeling of biological networks to analyzing social networks. Unveiling network dynamics allows us to make predictions and decisions. Moreover, network dynamics models have inspired new ideas for computational methods involving multi-agent cooperation, offering effective solutions for optimization tasks. This dissertation presents new theoretical results on network inference and multi-agent optimization, split into two parts -
The first part deals with modeling and identification of network dynamics. I study two types of network dynamics arising from social and gene networks. Based on the network dynamics, the proposed network identification method works like a `network RADAR', meaning that interaction strengths between agents are inferred by injecting `signal' into the network and observing the resultant reverberation. In social networks, this is accomplished by stubborn agents whose opinions do not change throughout a discussion. In gene networks, genes are suppressed to create desired perturbations. The steady-states under these perturbations are characterized. In contrast to the common assumption of full rank input, I take a laxer assumption where low-rank input is used, to better model the empirical network data. Importantly, a network is proven to be identifiable from low rank data of rank that grows proportional to the network's sparsity. The proposed method is applied to synthetic and empirical data, and is shown to offer superior performance compared to prior work. The second part is concerned with algorithms on networks. I develop three consensus-based algorithms for multi-agent optimization. The first method is a decentralized Frank-Wolfe (DeFW) algorithm. The main advantage of DeFW lies on its projection-free nature, where we can replace the costly projection step in traditional algorithms by a low-cost linear optimization step. I prove the convergence rates of DeFW for convex and non-convex problems. I also develop two consensus-based alternating optimization algorithms --- one for least square problems and one for non-convex problems. These algorithms exploit the problem structure for faster convergence and their efficacy is demonstrated by numerical simulations.
I conclude this dissertation by describing future research directions.
The first part deals with modeling and identification of network dynamics. I study two types of network dynamics arising from social and gene networks. Based on the network dynamics, the proposed network identification method works like a `network RADAR', meaning that interaction strengths between agents are inferred by injecting `signal' into the network and observing the resultant reverberation. In social networks, this is accomplished by stubborn agents whose opinions do not change throughout a discussion. In gene networks, genes are suppressed to create desired perturbations. The steady-states under these perturbations are characterized. In contrast to the common assumption of full rank input, I take a laxer assumption where low-rank input is used, to better model the empirical network data. Importantly, a network is proven to be identifiable from low rank data of rank that grows proportional to the network's sparsity. The proposed method is applied to synthetic and empirical data, and is shown to offer superior performance compared to prior work. The second part is concerned with algorithms on networks. I develop three consensus-based algorithms for multi-agent optimization. The first method is a decentralized Frank-Wolfe (DeFW) algorithm. The main advantage of DeFW lies on its projection-free nature, where we can replace the costly projection step in traditional algorithms by a low-cost linear optimization step. I prove the convergence rates of DeFW for convex and non-convex problems. I also develop two consensus-based alternating optimization algorithms --- one for least square problems and one for non-convex problems. These algorithms exploit the problem structure for faster convergence and their efficacy is demonstrated by numerical simulations.
I conclude this dissertation by describing future research directions.
ContributorsWai, Hoi To (Author) / Scaglione, Anna (Thesis advisor) / Berisha, Visar (Committee member) / Nedich, Angelia (Committee member) / Ying, Lei (Committee member) / Arizona State University (Publisher)
Created2017
Description
Design and mitigation of infrastructure on expansive soils requires an understanding of unsaturated soil mechanics and consideration of two stress variables (net normal stress and matric suction). Although numerous breakthroughs have allowed geotechnical engineers to study expansive soil response to varying suction-based stress scenarios (i.e. partial wetting), such studies are not practical on typical projects due to the difficulties and duration needed for equilibration associated with the necessary laboratory testing. The current practice encompasses saturated “conventional” soil mechanics testing, with the implementation of numerous empirical correlations and approximations to obtain an estimate of true field response. However, it has been observed that full wetting rarely occurs in the field, leading to an over-conservatism within a given design when partial wetting conditions are ignored. Many researchers have sought to improve ways of estimation of soil heave/shrinkage through intense studies of the suction-based response of reconstituted clay soils. However, the natural behavior of an undisturbed clay soil sample tends to differ significantly from a remolded sample of the same material.
In this study, laboratory techniques for the determination of soil suction were evaluated, a methodology for determination of the in-situ matric suction of a soil specimen was explored, and the mechanical response to changes in matric suction of natural clay specimens were measured. Suction-controlled laboratory oedometer devices were used to impose partial wetting conditions, similar to those experienced in a natural setting. The undisturbed natural soils tested in the study were obtained from Denver, CO and San Antonio, TX.
Key differences between the soil water characteristic curves of the undisturbed specimen test compared to the conventional reconstituted specimen test are highlighted. The Perko et al. (2000) and the PTI (2008) methods for estimating the relationship between volume and changes in matric suction (i.e. suction compression index) were evaluated by comparison to the directly measured values. Lastly, the directly measured partial wetting swell strain was compared to the fully saturated, one-dimensional, oedometer test (ASTM D4546) and the Surrogate Path Method (Singhal, 2010) to evaluate the estimation of partial wetting heave.
In this study, laboratory techniques for the determination of soil suction were evaluated, a methodology for determination of the in-situ matric suction of a soil specimen was explored, and the mechanical response to changes in matric suction of natural clay specimens were measured. Suction-controlled laboratory oedometer devices were used to impose partial wetting conditions, similar to those experienced in a natural setting. The undisturbed natural soils tested in the study were obtained from Denver, CO and San Antonio, TX.
Key differences between the soil water characteristic curves of the undisturbed specimen test compared to the conventional reconstituted specimen test are highlighted. The Perko et al. (2000) and the PTI (2008) methods for estimating the relationship between volume and changes in matric suction (i.e. suction compression index) were evaluated by comparison to the directly measured values. Lastly, the directly measured partial wetting swell strain was compared to the fully saturated, one-dimensional, oedometer test (ASTM D4546) and the Surrogate Path Method (Singhal, 2010) to evaluate the estimation of partial wetting heave.
ContributorsOlaiz, Austin Hunter (Author) / Houston, Sandra (Thesis advisor) / Zapata, Claudia (Committee member) / Kavazanjian, Edward (Committee member) / Arizona State University (Publisher)
Created2017
Description
This dissertation focuses on the application of urban metabolism metrology (UMM) to process streams of the natural and built water environment to gauge public health concerning exposure to carcinogenic N-nitrosamines and abuse of narcotics. A survey of sources of exposure to N-nitrosamines in the U.S. population identified contaminated food products (1,900 ± 380 ng/day) as important drivers of attributable cancer risk (Chapter 2). Freshwater sediments in the proximity of U.S. municipal wastewater treatment plants were shown for the first time to harbor carcinogenic N-nitrosamine congeners, including N-nitrosodibutylamine (0.2-3.3 ng/g dw), N-nitrosodiphenylamine (0.2-4.7 ng/g dw), and N-nitrosopyrrolidine (3.4-19.6 ng/g dw) were, with treated wastewater discharge representing one potential factor contributing to the observed contamination (p=0.42) (Chapter 3). Opioid abuse rates in two small midwestern communities were estimated through the application of wastewater-based epidemiology (WBE). Average concentrations of opioids (City 1; City 2) were highest for morphine (713 ± 38, 306 ± 29 ng/L) and varied by for the remainder of the screened analytes. Furthermore, concentrations of the powerful opioid fentanyl (1.7 ± 0.2, 1.0 ± 0.5 ng/L) in wastewater were reported for the first time in the literature for the U.S. (Chapter 4). To gauge narcotic consumption within college-aged adults the WBE process used in Chapter 4 was applied to wastewater collected from a large university in the Southwestern U.S. Estimated narcotics consumption, in units of mg/day/1,000 persons showed the following rank order: cocaine (470 ± 42), heroin (474 ± 32), amphetamine (302 ± 14) and methylphenidate (236 ± 28). Most parental drugs and their respective metabolites showed detection frequencies in campus wastewater of 80% or more, with the notable exception of fentanyl, norfentanyl, buprenorphine, and norbuprenorphine. Estimated consumption of all narcotics, aside from attention-deficit/hyperactivity disorder medication, were higher than values reported in previous U.S. WBE studies for U.S. campuses (Chapter 5). The analyses presented here have identified variation in narcotic consumption habits across different U.S. communities, which can be gauged through UMM. Application of these techniques should be implemented throughout U.S. communities to provide insight into ongoing substance abuse and health issues within a community.
ContributorsGushgari, Adam Jon (Author) / Halden, Rolf U. (Thesis advisor) / Kavazanjian, Edward (Committee member) / Fraser, Matthew (Committee member) / Venkatesan, Arjun (Committee member) / Mascaro, Giuseppe (Committee member) / Arizona State University (Publisher)
Created2018
Description
The activation of the primary motor cortex (M1) is common in speech perception tasks that involve difficult listening conditions. Although the challenge of recognizing and discriminating non-native speech sounds appears to be an instantiation of listening under difficult circumstances, it is still unknown if M1 recruitment is facilitatory of second language speech perception. The purpose of this study was to investigate the role of M1 associated with speech motor centers in processing acoustic inputs in the native (L1) and second language (L2), using repetitive Transcranial Magnetic Stimulation (rTMS) to selectively alter neural activity in M1. Thirty-six healthy English/Spanish bilingual subjects participated in the experiment. The performance on a listening word-to-picture matching task was measured before and after real- and sham-rTMS to the orbicularis oris (lip muscle) associated M1. Vowel Space Area (VSA) obtained from recordings of participants reading a passage in L2 before and after real-rTMS, was calculated to determine its utility as an rTMS aftereffect measure. There was high variability in the aftereffect of the rTMS protocol to the lip muscle among the participants. Approximately 50% of participants showed an inhibitory effect of rTMS, evidenced by smaller motor evoked potentials (MEPs) area, whereas the other 50% had a facilitatory effect, with larger MEPs. This suggests that rTMS has a complex influence on M1 excitability, and relying on grand-average results can obscure important individual differences in rTMS physiological and functional outcomes. Evidence of motor support to word recognition in the L2 was found. Participants showing an inhibitory aftereffect of rTMS on M1 produced slower and less accurate responses in the L2 task, whereas those showing a facilitatory aftereffect of rTMS on M1 produced more accurate responses in L2. In contrast, no effect of rTMS was found on the L1, where accuracy and speed were very similar after sham- and real-rTMS. The L2 VSA measure was indicative of the aftereffect of rTMS to M1 associated with speech production, supporting its utility as an rTMS aftereffect measure. This result revealed an interesting and novel relation between cerebral motor cortex activation and speech measures.
ContributorsBarragan, Beatriz (Author) / Liss, Julie (Thesis advisor) / Berisha, Visar (Committee member) / Rogalsky, Corianne (Committee member) / Restrepo, Adelaida (Committee member) / Arizona State University (Publisher)
Created2018
Description
Modern machine learning systems leverage data and features from multiple modalities to gain more predictive power. In most scenarios, the modalities are vastly different and the acquired data are heterogeneous in nature. Consequently, building highly effective fusion algorithms is at the core to achieve improved model robustness and inferencing performance. This dissertation focuses on the representation learning approaches as the fusion strategy. Specifically, the objective is to learn the shared latent representation which jointly exploit the structural information encoded in all modalities, such that a straightforward learning model can be adopted to obtain the prediction.
We first consider sensor fusion, a typical multimodal fusion problem critical to building a pervasive computing platform. A systematic fusion technique is described to support both multiple sensors and descriptors for activity recognition. Targeted to learn the optimal combination of kernels, Multiple Kernel Learning (MKL) algorithms have been successfully applied to numerous fusion problems in computer vision etc. Utilizing the MKL formulation, next we describe an auto-context algorithm for learning image context via the fusion with low-level descriptors. Furthermore, a principled fusion algorithm using deep learning to optimize kernel machines is developed. By bridging deep architectures with kernel optimization, this approach leverages the benefits of both paradigms and is applied to a wide variety of fusion problems.
In many real-world applications, the modalities exhibit highly specific data structures, such as time sequences and graphs, and consequently, special design of the learning architecture is needed. In order to improve the temporal modeling for multivariate sequences, we developed two architectures centered around attention models. A novel clinical time series analysis model is proposed for several critical problems in healthcare. Another model coupled with triplet ranking loss as metric learning framework is described to better solve speaker diarization. Compared to state-of-the-art recurrent networks, these attention-based multivariate analysis tools achieve improved performance while having a lower computational complexity. Finally, in order to perform community detection on multilayer graphs, a fusion algorithm is described to derive node embedding from word embedding techniques and also exploit the complementary relational information contained in each layer of the graph.
We first consider sensor fusion, a typical multimodal fusion problem critical to building a pervasive computing platform. A systematic fusion technique is described to support both multiple sensors and descriptors for activity recognition. Targeted to learn the optimal combination of kernels, Multiple Kernel Learning (MKL) algorithms have been successfully applied to numerous fusion problems in computer vision etc. Utilizing the MKL formulation, next we describe an auto-context algorithm for learning image context via the fusion with low-level descriptors. Furthermore, a principled fusion algorithm using deep learning to optimize kernel machines is developed. By bridging deep architectures with kernel optimization, this approach leverages the benefits of both paradigms and is applied to a wide variety of fusion problems.
In many real-world applications, the modalities exhibit highly specific data structures, such as time sequences and graphs, and consequently, special design of the learning architecture is needed. In order to improve the temporal modeling for multivariate sequences, we developed two architectures centered around attention models. A novel clinical time series analysis model is proposed for several critical problems in healthcare. Another model coupled with triplet ranking loss as metric learning framework is described to better solve speaker diarization. Compared to state-of-the-art recurrent networks, these attention-based multivariate analysis tools achieve improved performance while having a lower computational complexity. Finally, in order to perform community detection on multilayer graphs, a fusion algorithm is described to derive node embedding from word embedding techniques and also exploit the complementary relational information contained in each layer of the graph.
ContributorsSong, Huan (Author) / Spanias, Andreas (Thesis advisor) / Thiagarajan, Jayaraman (Committee member) / Berisha, Visar (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Arizona State University (Publisher)
Created2018
Description
This dissertation is focused on developing an algorithm to provide current state estimation and future state predictions for biomechanical human walking features. The goal is to develop a system which is capable of evaluating the current action a subject is taking while walking and then use this to predict the future states of biomechanical features.
This work focuses on the exploration and analysis of Interaction Primitives (Amor er al, 2014) and their relevance to biomechanical prediction for human walking. Built on the framework of Probabilistic Movement Primitives, Interaction Primitives utilize an EKF SLAM algorithm to localize and map a distribution over the weights of a set of basis functions. The prediction properties of Bayesian Interaction Primitives were utilized to predict real-time foot forces from a 9 degrees of freedom IMUs mounted to a subjects tibias. This method shows that real-time human biomechanical features can be predicted and have a promising link to real-time controls applications.
This work focuses on the exploration and analysis of Interaction Primitives (Amor er al, 2014) and their relevance to biomechanical prediction for human walking. Built on the framework of Probabilistic Movement Primitives, Interaction Primitives utilize an EKF SLAM algorithm to localize and map a distribution over the weights of a set of basis functions. The prediction properties of Bayesian Interaction Primitives were utilized to predict real-time foot forces from a 9 degrees of freedom IMUs mounted to a subjects tibias. This method shows that real-time human biomechanical features can be predicted and have a promising link to real-time controls applications.
ContributorsClark, Geoffrey Mitchell (Author) / Ben Amor, Heni (Thesis advisor) / Si, Jennie (Committee member) / Berisha, Visar (Committee member) / Arizona State University (Publisher)
Created2018
Description
Deep neural networks (DNN) have shown tremendous success in various cognitive tasks, such as image classification, speech recognition, etc. However, their usage on resource-constrained edge devices has been limited due to high computation and large memory requirement.
To overcome these challenges, recent works have extensively investigated model compression techniques such as element-wise sparsity, structured sparsity and quantization. While most of these works have applied these compression techniques in isolation, there have been very few studies on application of quantization and structured sparsity together on a DNN model.
This thesis co-optimizes structured sparsity and quantization constraints on DNN models during training. Specifically, it obtains optimal setting of 2-bit weight and 2-bit activation coupled with 4X structured compression by performing combined exploration of quantization and structured compression settings. The optimal DNN model achieves 50X weight memory reduction compared to floating-point uncompressed DNN. This memory saving is significant since applying only structured sparsity constraints achieves 2X memory savings and only quantization constraints achieves 16X memory savings. The algorithm has been validated on both high and low capacity DNNs and on wide-sparse and deep-sparse DNN models. Experiments demonstrated that deep-sparse DNN outperforms shallow-dense DNN with varying level of memory savings depending on DNN precision and sparsity levels. This work further proposed a Pareto-optimal approach to systematically extract optimal DNN models from a huge set of sparse and dense DNN models. The resulting 11 optimal designs were further evaluated by considering overall DNN memory which includes activation memory and weight memory. It was found that there is only a small change in the memory footprint of the optimal designs corresponding to the low sparsity DNNs. However, activation memory cannot be ignored for high sparsity DNNs.
To overcome these challenges, recent works have extensively investigated model compression techniques such as element-wise sparsity, structured sparsity and quantization. While most of these works have applied these compression techniques in isolation, there have been very few studies on application of quantization and structured sparsity together on a DNN model.
This thesis co-optimizes structured sparsity and quantization constraints on DNN models during training. Specifically, it obtains optimal setting of 2-bit weight and 2-bit activation coupled with 4X structured compression by performing combined exploration of quantization and structured compression settings. The optimal DNN model achieves 50X weight memory reduction compared to floating-point uncompressed DNN. This memory saving is significant since applying only structured sparsity constraints achieves 2X memory savings and only quantization constraints achieves 16X memory savings. The algorithm has been validated on both high and low capacity DNNs and on wide-sparse and deep-sparse DNN models. Experiments demonstrated that deep-sparse DNN outperforms shallow-dense DNN with varying level of memory savings depending on DNN precision and sparsity levels. This work further proposed a Pareto-optimal approach to systematically extract optimal DNN models from a huge set of sparse and dense DNN models. The resulting 11 optimal designs were further evaluated by considering overall DNN memory which includes activation memory and weight memory. It was found that there is only a small change in the memory footprint of the optimal designs corresponding to the low sparsity DNNs. However, activation memory cannot be ignored for high sparsity DNNs.
ContributorsSrivastava, Gaurav (Author) / Seo, Jae-Sun (Thesis advisor) / Chakrabarti, Chaitali (Committee member) / Berisha, Visar (Committee member) / Arizona State University (Publisher)
Created2018