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 31 - 40 of 316
Filtering by
- Creators: School of Historical, Philosophical and Religious Studies
- Creators: Turaga, Pavan
Description
Mixture of experts is a machine learning ensemble approach that consists of individual models that are trained to be ``experts'' on subsets of the data, and a gating network that provides weights to output a combination of the expert predictions. Mixture of experts models do not currently see wide use due to difficulty in training diverse experts and high computational requirements. This work presents modifications of the mixture of experts formulation that use domain knowledge to improve training, and incorporate parameter sharing among experts to reduce computational requirements.
First, this work presents an application of mixture of experts models for quality robust visual recognition. First it is shown that human subjects outperform deep neural networks on classification of distorted images, and then propose a model, MixQualNet, that is more robust to distortions. The proposed model consists of ``experts'' that are trained on a particular type of image distortion. The final output of the model is a weighted sum of the expert models, where the weights are determined by a separate gating network. The proposed model also incorporates weight sharing to reduce the number of parameters, as well as increase performance.
Second, an application of mixture of experts to predict visual saliency is presented. A computational saliency model attempts to predict where humans will look in an image. In the proposed model, each expert network is trained to predict saliency for a set of closely related images. The final saliency map is computed as a weighted mixture of the expert networks' outputs, with weights determined by a separate gating network. The proposed model achieves better performance than several other visual saliency models and a baseline non-mixture model.
Finally, this work introduces a saliency model that is a weighted mixture of models trained for different levels of saliency. Levels of saliency include high saliency, which corresponds to regions where almost all subjects look, and low saliency, which corresponds to regions where some, but not all subjects look. The weighted mixture shows improved performance compared with baseline models because of the diversity of the individual model predictions.
First, this work presents an application of mixture of experts models for quality robust visual recognition. First it is shown that human subjects outperform deep neural networks on classification of distorted images, and then propose a model, MixQualNet, that is more robust to distortions. The proposed model consists of ``experts'' that are trained on a particular type of image distortion. The final output of the model is a weighted sum of the expert models, where the weights are determined by a separate gating network. The proposed model also incorporates weight sharing to reduce the number of parameters, as well as increase performance.
Second, an application of mixture of experts to predict visual saliency is presented. A computational saliency model attempts to predict where humans will look in an image. In the proposed model, each expert network is trained to predict saliency for a set of closely related images. The final saliency map is computed as a weighted mixture of the expert networks' outputs, with weights determined by a separate gating network. The proposed model achieves better performance than several other visual saliency models and a baseline non-mixture model.
Finally, this work introduces a saliency model that is a weighted mixture of models trained for different levels of saliency. Levels of saliency include high saliency, which corresponds to regions where almost all subjects look, and low saliency, which corresponds to regions where some, but not all subjects look. The weighted mixture shows improved performance compared with baseline models because of the diversity of the individual model predictions.
ContributorsDodge, Samuel Fuller (Author) / Karam, Lina (Thesis advisor) / Jayasuriya, Suren (Committee member) / Li, Baoxin (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2018
Description
Motion estimation is a core task in computer vision and many applications utilize optical flow methods as fundamental tools to analyze motion in images and videos. Optical flow is the apparent motion of objects in image sequences that results from relative motion between the objects and the imaging perspective. Today, optical flow fields are utilized to solve problems in various areas such as object detection and tracking, interpolation, visual odometry, etc. In this dissertation, three problems from different areas of computer vision and the solutions that make use of modified optical flow methods are explained.
The contributions of this dissertation are approaches and frameworks that introduce i) a new optical flow-based interpolation method to achieve minimally divergent velocimetry data, ii) a framework that improves the accuracy of change detection algorithms in synthetic aperture radar (SAR) images, and iii) a set of new methods to integrate Proton Magnetic Resonance Spectroscopy (1HMRSI) data into threedimensional (3D) neuronavigation systems for tumor biopsies.
In the first application an optical flow-based approach for the interpolation of minimally divergent velocimetry data is proposed. The velocimetry data of incompressible fluids contain signals that describe the flow velocity. The approach uses the additional flow velocity information to guide the interpolation process towards reduced divergence in the interpolated data.
In the second application a framework that mainly consists of optical flow methods and other image processing and computer vision techniques to improve object extraction from synthetic aperture radar images is proposed. The proposed framework is used for distinguishing between actual motion and detected motion due to misregistration in SAR image sets and it can lead to more accurate and meaningful change detection and improve object extraction from a SAR datasets.
In the third application a set of new methods that aim to improve upon the current state-of-the-art in neuronavigation through the use of detailed three-dimensional (3D) 1H-MRSI data are proposed. The result is a progressive form of online MRSI-guided neuronavigation that is demonstrated through phantom validation and clinical application.
The contributions of this dissertation are approaches and frameworks that introduce i) a new optical flow-based interpolation method to achieve minimally divergent velocimetry data, ii) a framework that improves the accuracy of change detection algorithms in synthetic aperture radar (SAR) images, and iii) a set of new methods to integrate Proton Magnetic Resonance Spectroscopy (1HMRSI) data into threedimensional (3D) neuronavigation systems for tumor biopsies.
In the first application an optical flow-based approach for the interpolation of minimally divergent velocimetry data is proposed. The velocimetry data of incompressible fluids contain signals that describe the flow velocity. The approach uses the additional flow velocity information to guide the interpolation process towards reduced divergence in the interpolated data.
In the second application a framework that mainly consists of optical flow methods and other image processing and computer vision techniques to improve object extraction from synthetic aperture radar images is proposed. The proposed framework is used for distinguishing between actual motion and detected motion due to misregistration in SAR image sets and it can lead to more accurate and meaningful change detection and improve object extraction from a SAR datasets.
In the third application a set of new methods that aim to improve upon the current state-of-the-art in neuronavigation through the use of detailed three-dimensional (3D) 1H-MRSI data are proposed. The result is a progressive form of online MRSI-guided neuronavigation that is demonstrated through phantom validation and clinical application.
ContributorsKanberoglu, Berkay (Author) / Frakes, David (Thesis advisor) / Turaga, Pavan (Thesis advisor) / Spanias, Andreas (Committee member) / Berisha, Visar (Committee member) / Arizona State University (Publisher)
Created2018
Description
Human movement is a complex process influenced by physiological and psychological factors. The execution of movement is varied from person to person, and the number of possible strategies for completing a specific movement task is almost infinite. Different choices of strategies can be perceived by humans as having different degrees of quality, and the quality can be defined with regard to aesthetic, athletic, or health-related ratings. It is useful to measure and track the quality of a person's movements, for various applications, especially with the prevalence of low-cost and portable cameras and sensors today. Furthermore, based on such measurements, feedback systems can be designed for people to practice their movements towards certain goals. In this dissertation, I introduce symmetry as a family of measures for movement quality, and utilize recent advances in computer vision and differential geometry to model and analyze different types of symmetry in human movements. Movements are modeled as trajectories on different types of manifolds, according to the representations of movements from sensor data. The benefit of such a universal framework is that it can accommodate different existing and future features that describe human movements. The theory and tools developed in this dissertation will also be useful in other scientific areas to analyze symmetry from high-dimensional signals.
ContributorsWang, Qiao (Author) / Turaga, Pavan (Thesis advisor) / Spanias, Andreas (Committee member) / Srivastava, Anuj (Committee member) / Sha, Xin Wei (Committee member) / Arizona State University (Publisher)
Created2018
Description
Non-line-of-sight (NLOS) imaging of objects not visible to either the camera or illumina-
tion source is a challenging task with vital applications including surveillance and robotics.
Recent NLOS reconstruction advances have been achieved using time-resolved measure-
ments. Acquiring these time-resolved measurements requires expensive and specialized
detectors and laser sources. In work proposes a data-driven approach for NLOS 3D local-
ization requiring only a conventional camera and projector. The localisation is performed
using a voxelisation and a regression problem. Accuracy of greater than 90% is achieved
in localizing a NLOS object to a 5cm × 5cm × 5cm volume in real data. By adopting
the regression approach an object of width 10cm to localised to approximately 1.5cm. To
generalize to line-of-sight (LOS) scenes with non-planar surfaces, an adaptive lighting al-
gorithm is adopted. This algorithm, based on radiosity, identifies and illuminates scene
patches in the LOS which most contribute to the NLOS light paths, and can factor in sys-
tem power constraints. Improvements ranging from 6%-15% in accuracy with a non-planar
LOS wall using adaptive lighting is reported, demonstrating the advantage of combining
the physics of light transport with active illumination for data-driven NLOS imaging.
tion source is a challenging task with vital applications including surveillance and robotics.
Recent NLOS reconstruction advances have been achieved using time-resolved measure-
ments. Acquiring these time-resolved measurements requires expensive and specialized
detectors and laser sources. In work proposes a data-driven approach for NLOS 3D local-
ization requiring only a conventional camera and projector. The localisation is performed
using a voxelisation and a regression problem. Accuracy of greater than 90% is achieved
in localizing a NLOS object to a 5cm × 5cm × 5cm volume in real data. By adopting
the regression approach an object of width 10cm to localised to approximately 1.5cm. To
generalize to line-of-sight (LOS) scenes with non-planar surfaces, an adaptive lighting al-
gorithm is adopted. This algorithm, based on radiosity, identifies and illuminates scene
patches in the LOS which most contribute to the NLOS light paths, and can factor in sys-
tem power constraints. Improvements ranging from 6%-15% in accuracy with a non-planar
LOS wall using adaptive lighting is reported, demonstrating the advantage of combining
the physics of light transport with active illumination for data-driven NLOS imaging.
ContributorsChandran, Sreenithy (Author) / Jayasuriya, Suren (Thesis advisor) / Turaga, Pavan (Committee member) / Dasarathy, Gautam (Committee member) / Arizona State University (Publisher)
Created2019
Description
Speech is generated by articulators acting on
a phonatory source. Identification of this
phonatory source and articulatory geometry are
individually challenging and ill-posed
problems, called speech separation and
articulatory inversion, respectively.
There exists a trade-off
between decomposition and recovered
articulatory geometry due to multiple
possible mappings between an
articulatory configuration
and the speech produced. However, if measurements
are obtained only from a microphone sensor,
they lack any invasive insight and add
additional challenge to an already difficult
problem.
A joint non-invasive estimation
strategy that couples articulatory and
phonatory knowledge would lead to better
articulatory speech synthesis. In this thesis,
a joint estimation strategy for speech
separation and articulatory geometry recovery
is studied. Unlike previous
periodic/aperiodic decomposition methods that
use stationary speech models within a
frame, the proposed model presents a
non-stationary speech decomposition method.
A parametric glottal source model and an
articulatory vocal tract response are
represented in a dynamic state space formulation.
The unknown parameters of the
speech generation components are estimated
using sequential Monte Carlo methods
under some specific assumptions.
The proposed approach is compared with other
glottal inverse filtering methods,
including iterative adaptive inverse filtering,
state-space inverse filtering, and
the quasi-closed phase method.
a phonatory source. Identification of this
phonatory source and articulatory geometry are
individually challenging and ill-posed
problems, called speech separation and
articulatory inversion, respectively.
There exists a trade-off
between decomposition and recovered
articulatory geometry due to multiple
possible mappings between an
articulatory configuration
and the speech produced. However, if measurements
are obtained only from a microphone sensor,
they lack any invasive insight and add
additional challenge to an already difficult
problem.
A joint non-invasive estimation
strategy that couples articulatory and
phonatory knowledge would lead to better
articulatory speech synthesis. In this thesis,
a joint estimation strategy for speech
separation and articulatory geometry recovery
is studied. Unlike previous
periodic/aperiodic decomposition methods that
use stationary speech models within a
frame, the proposed model presents a
non-stationary speech decomposition method.
A parametric glottal source model and an
articulatory vocal tract response are
represented in a dynamic state space formulation.
The unknown parameters of the
speech generation components are estimated
using sequential Monte Carlo methods
under some specific assumptions.
The proposed approach is compared with other
glottal inverse filtering methods,
including iterative adaptive inverse filtering,
state-space inverse filtering, and
the quasi-closed phase method.
ContributorsVenkataramani, Adarsh Akkshai (Author) / Papandreou-Suppappola, Antonia (Thesis advisor) / Bliss, Daniel W (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2018
Description
At present, the vast majority of human subjects with neurological disease are still diagnosed through in-person assessments and qualitative analysis of patient data. In this paper, we propose to use Topological Data Analysis (TDA) together with machine learning tools to automate the process of Parkinson’s disease classification and severity assessment. An automated, stable, and accurate method to evaluate Parkinson’s would be significant in streamlining diagnoses of patients and providing families more time for corrective measures. We propose a methodology which incorporates TDA into analyzing Parkinson’s disease postural shifts data through the representation of persistence images. Studying the topology of a system has proven to be invariant to small changes in data and has been shown to perform well in discrimination tasks. The contributions of the paper are twofold. We propose a method to 1) classify healthy patients from those afflicted by disease and 2) diagnose the severity of disease. We explore the use of the proposed method in an application involving a Parkinson’s disease dataset comprised of healthy-elderly, healthy-young and Parkinson’s disease patients.
ContributorsRahman, Farhan Nadir (Co-author) / Nawar, Afra (Co-author) / Turaga, Pavan (Thesis director) / Krishnamurthi, Narayanan (Committee member) / Electrical Engineering Program (Contributor) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
Description
During the 1970's to 90's, scholars in the fields of Jewish Studies, anthropology and sociology (notably, Hellen Epstein, Larry Langer and Yosef Yerushalmi), developed the idea of generational trauma theory, when analyzing the trauma inflicted upon European Jewish populations during the Holocaust. Epstein argues that trauma is passed down from generation to generation, while Langer argues that the second generation interprets the trauma in their own way. Other important terms in trauma theory include liturgical time, sites of memory, historical trauma and the historical trauma response. Scholars who analyze American Indian communities, like Yellow Horse Brave Heart and Durran/Durran, readily took up this theory, applying it to the Native American experience. One area where this theory has been applied to is the Native American Boarding School experience. The purpose of this paper is to analyze the efficacy of applying the post-Holocaust trauma theory to the Native American boarding school experience. In order to determine the effectiveness of the boarding schools, one must analyze the boarding school experience, beginning at the philosophical underpinnings of the boarding school, and then discussing the impacts that the boarding schools had on the students and finally, the impact that this had on the second generation. However, this approach has a number of flaws, such as the differences between native communities and post-Holocaust, American, Jewish communities, as discussed in the Philosophy of American Indian Studies. The length of the boarding schools was also longer than the length of the Holocaust. The fact that Native Americans faced repeated trauma, in a way that post-Holocaust American Jews did not. The trauma also changed for both native peoples and post-Holocaust Jews, making it difficult for there to be a single response to trauma. The philosophical bases of the Holocaust and boarding schools were also different. The post-Holocaust generational trauma approach also has a number of applications to native peoples. This includes the psychological aspect of trauma. The use of terminology by native scholars. Native peoples also developed concepts like sites of memory and liturgical time. Finally, both the post-Holocaust Jewas and Native Americans have used trauma for political ends. The conclusion is that post-Holocaust generational trauma theory has some applications to native peoples, but the application is limited. A scholar must take into careful consideration the native peoples who they are working with.
ContributorsMongeau, Michael Philip (Author) / Benkert, Volker (Thesis director) / Riding In, James (Committee member) / School of Historical, Philosophical and Religious Studies (Contributor) / American Indian Studies Program (Contributor) / Barrett, The Honors College (Contributor)
Created2017-12
Description
Sigmund Freud's psychoanalytic theory proposes that the personality has three components, the id, superego, and ego. The id is concerned with pleasure and gain, the reason it is often identified as a human's animalistic side. Additionally, the id does not consider social rules as closely and is the uncensored portion of the personality. The superego is the id's opposite; the superego considers social expectations and pressures immensely, is more self-critical and moralizing. The ego mediates the id and superego, and is understood as the realistic expression of personality which considers both the "animal" and human. A Fractured Whole: A Collection of Short Stories, explores Freud's construction of human personality in both form and content. Within the collection are three sections, each with a different pair of characters. Within each section, the same scene is written in the three "modes" of the id, superego, and ego, as three separate stories. The fifteen stories comprising this collection address the substance of daily life: sexuality, body image, competition, among other topics, to consider how a single person can balance the desires for personal pleasure and to satisfy social expectations. Writing the same scene in three "modes" allows for the observation of how the characters attitudes and actions alter under the influence of different parts of their personalities.
ContributorsOtte, Aneka (Author) / Sturges, Robert (Thesis director) / Bryant, Jason (Committee member) / School of Historical, Philosophical and Religious Studies (Contributor) / Department of English (Contributor) / Barrett, The Honors College (Contributor)
Created2016-05
Description
Since Dylan Roof, a white supremacist, shot and killed nine members of a black church in Charleston on June 17, 2015, Confederate symbols have stood at the center of much controversy across the United States. Although the Confederate battle flag remains the most obvious example, the debate took a particular form in Tennessee, centering on the image of General Nathan Bedford Forrest. Born in 1822 to a poor family, he left school early to work. Although his work in the slave trade made him a millionaire, his later participation in the massacre of over 300 black soldiers at Fort Pillow in 1864 during the Civil War and association with the Ku Klux Klan cemented his reputation as a violent racist. Yet, many white Tennesseans praised him as a hero and memorialized him. This thesis examines Nathan Bedford Forrest State Park in Benton County and Forrest Park, now Health Sciences Park, in Memphis to examine what characteristics denote a controversial memorial. Specifically, I focus on the physical form, the location, and the demographics of the area, investigating how these components work together to give rise to controversy or acceptance of the memorial's image. Physical representations greatly impact the ideas associated with the memorial while racial demographics affect whether or not Forrest's representation as a hero speaks true to modern interpretations and opinions.
ContributorsFish, Caroline Rose (Author) / Longley, Rodney (Thesis director) / Suk, Mina (Committee member) / School of Historical, Philosophical and Religious Studies (Contributor) / School of Politics and Global Studies (Contributor) / School of International Letters and Cultures (Contributor) / Barrett, The Honors College (Contributor)
Created2016-05
Description
This thesis explores the intersection of religion, social class, and politics during the late nineteenth century in Imperial Germany. Specifically, the focus of this work is on the Workers' Association of Saint Paul's in Aachen and Burtscheid, a Catholic working-class organization in the 1870s located in the city of Aachen, a rapidly industrializing city in the majority Catholic Prussian Rhineland. This organization was the largest Catholic working-class association in Germany in the 1870s, reaching 5,000 members by the middle of the decade, and also espoused the politics of Christian Socialism. This thesis explores the intersection of the possibly competing social identities of these workers between being Catholics and workers. To start, the scholarly framework of studying society and politics in Imperial Germany is discussed, especially the concept of rigidly constructed social milieus into five groups, two of them being the Catholics and the working-class, and how this work may suggest a challenge to this concept. Next, the background information of how a Catholic working-class came into existence, as it was the product of simultaneous nineteenth century processes of industrialization and a religious revival among German Catholics. The Kulturkampf was the force that politicized Catholicism in Germany, as the persecution of Catholic institutions by Prussia forced Catholics into a social and political "ghetto." Then, the Association of St. Paul's itself is discussed. First, the workers espoused their Catholic identity and religious solidarity during a time of persecution, but also emphasized the Christian basis for their brand of Socialism. This lent into their identity as part of the working-class. While they steadfastly rejected the "godlessness" of Social Democracy, the Christian Socialists also shared many similar social and political goals. This intersection between identities eventually led to political conflict in Aachen throughout the 1870s with the mainstream, bourgeois Catholics of the city. To conclude, the legacy of Christian Socialism on modern Germany is discussed, as well as its contribution to the complex politics of Imperial Germany.
ContributorsHarkness, John Hosea (Author) / Benkert, Volker (Thesis director) / Holian, Anna (Committee member) / School of Politics and Global Studies (Contributor) / School of International Letters and Cultures (Contributor) / School of Historical, Philosophical and Religious Studies (Contributor) / Barrett, The Honors College (Contributor)
Created2016-05