Matching Items (78)
Description
Graph matching is a fundamental but notoriously difficult problem due to its NP-hard nature, and serves as a cornerstone for a series of applications in machine learning and computer vision, such as image matching, dynamic routing, drug design, to name a few. Although there has been massive previous investigation on high-performance graph matching solvers, it still remains a challenging task to tackle the matching problem under real-world scenarios with severe graph uncertainty (e.g., noise, outlier, misleading or ambiguous link).In this dissertation, a main focus is to investigate the essence and propose solutions to graph matching with higher reliability under such uncertainty. To this end, the proposed research was conducted taking into account three perspectives related to reliable graph matching: modeling, optimization and learning. For modeling, graph matching is extended from typical quadratic assignment problem to a more generic mathematical model by introducing a specific family of separable function, achieving higher capacity and reliability. In terms of optimization, a novel high gradient-efficient determinant-based regularization technique is proposed in this research, showing high robustness against outliers. Then learning paradigm for graph matching under intrinsic combinatorial characteristics is explored. First, a study is conducted on the way of filling the gap between discrete problem and its continuous approximation under a deep learning framework. Then this dissertation continues to investigate the necessity of more reliable latent topology of graphs for matching, and propose an effective and flexible framework to obtain it. Coherent findings in this dissertation include theoretical study and several novel algorithms, with rich experiments demonstrating the effectiveness.
ContributorsYu, Tianshu (Author) / Li, Baoxin (Thesis advisor) / Wang, Yalin (Committee member) / Yang, Yezhou (Committee member) / Yang, Yingzhen (Committee member) / Arizona State University (Publisher)
Created2021
Description
Graph matching is a fundamental but notoriously difficult problem due to its NP-hard nature, and serves as a cornerstone for a series of applications in machine learning and computer vision, such as image matching, dynamic routing, drug design, to name a few. Although there has been massive previous investigation on high-performance graph matching solvers, it still remains a challenging task to tackle the matching problem under real-world scenarios with severe graph uncertainty (e.g., noise, outlier, misleading or ambiguous link).In this dissertation, a main focus is to investigate the essence and propose solutions to graph matching with higher reliability under such uncertainty. To this end, the proposed research was conducted taking into account three perspectives related to reliable graph matching: modeling, optimization and learning. For modeling, graph matching is extended from typical quadratic assignment problem to a more generic mathematical model by introducing a specific family of separable function, achieving higher capacity and reliability. In terms of optimization, a novel high gradient-efficient determinant-based regularization technique is proposed in this research, showing high robustness against outliers. Then learning paradigm for graph matching under intrinsic combinatorial characteristics is explored. First, a study is conducted on the way of filling the gap between discrete problem and its continuous approximation under a deep learning framework. Then this dissertation continues to investigate the necessity of more reliable latent topology of graphs for matching, and propose an effective and flexible framework to obtain it. Coherent findings in this dissertation include theoretical study and several novel algorithms, with rich experiments demonstrating the effectiveness.
ContributorsYu, Tianshu (Author) / Li, Baoxin (Thesis advisor) / Wang, Yalin (Committee member) / Yang, Yezhou (Committee member) / Yang, Yingzhen (Committee member) / Arizona State University (Publisher)
Created2021
Description
Alzheimer's disease (AD) is a neurodegenerative disease that damages the cognitive abilities of a patient. It is critical to diagnose AD early to begin treatment as soon as possible which can be done through biomarkers. One such biomarker is the beta-amyloid (Aβ) peptide which can be quantified using the centiloid (CL) scale. For identifying the Aβ biomarker, A deep learning model that can model AD progression by predicting the CL value for brain magnetic resonance images (MRIs) is proposed. Brain MRI images can be obtained through the Alzheimer's Disease Neuroimaging Initiative (ADNI) and Open Access Series of Imaging Studies (OASIS) datasets, however a single model cannot perform well on both datasets at once. Thus, A regularization-based continuous learning framework to perform domain adaptation on the previous model is also proposed which captures the latent information about the relationship between Aβ and AD progression within both datasets.
ContributorsTrinh, Matthew Brian (Author) / Wang, Yalin (Thesis advisor) / Liang, Jianming (Committee member) / Su, Yi (Committee member) / Arizona State University (Publisher)
Created2022
Description
The Population Receptive Field (pRF) model is widely used to predict the location (retinotopy) and size of receptive fields on the visual space. Doing so allows for the creation of a mapping from locations in the visual field to the associated groups of neurons in the cortical region (within the visual cortex of the brain). However, using the pRF model is very time consuming. Past research has focused on the creation of Convolutional Neural Networks (CNN) to mimic the pRF model in a fraction of the time, and they have worked well under highly controlled conditions. However, these models have not been thoroughly tested on real human data. This thesis focused on adapting one of these CNNs to accurately predict the retinotopy of a real human subject using a dataset from the Human Connectome Project. The results show promise towards creating a fully functioning CNN, but they also expose new challenges that must be overcome before the model could be used to predict the retinotopy of new human subjects.
ContributorsBurgard, Braeden (Author) / Wang, Yalin (Thesis director) / Ta, Duyan (Committee member) / Barrett, The Honors College (Contributor) / School of International Letters and Cultures (Contributor) / Computer Science and Engineering Program (Contributor) / School of Mathematical and Statistical Sciences (Contributor)
Created2022-05
Description
Attitudes and habits are extremely resistant to change, but a disruption of the magnitude of the COVID-19 pandemic has the potential to bring long-term, massive societal changes. During the pandemic, people are being compelled to experience new ways of interacting, working, learning, shopping, traveling, and eating meals. Going forward, a critical question is whether these experiences will result in changed behaviors and preferences in the long term. This paper presents initial findings on the likelihood of long-term changes in telework, daily travel, restaurant patronage, and air travel based on survey data collected from adults in the United States in Spring 2020. These data suggest that a sizable fraction of the increase in telework and decreases in both business air travel and restaurant patronage are likely here to stay. As for daily travel modes, public transit may not fully recover its pre-pandemic ridership levels, but many of our respondents are planning to bike and walk more than they used to. These data reflect the responses of a sample that is higher income and more highly educated than the US population. The response of these particular groups to the COVID-19 pandemic is perhaps especially important to understand, however, because their consumption patterns give them a large influence on many sectors of the economy.
ContributorsConway, Matthew Wigginton (Author) / Salon, Deborah (Author) / da Silva, Denise Capasso (Author) / Mirtich, Laura Christine (Author)
Created2020-09-03
Description
The purpose of the overall project is to create a simulated environment similar to Google map and traffic but simplified for education purposes. Students can choose different traffic patterns and program a car to navigate through the traffic dynamically based on the changing traffic. The environment used in the project is ASU VIPLE (Visual IoT/Robotics Programming Language Environment). It is a visual programming environment for Computer Science education. VIPLE supports a number of devices and platforms, including a traffic simulator developed using Unity game engine. This thesis focuses on creating realistic traffic data for the traffic simulator and implementing dynamic routing algorithm in VIPLE. The traffic data is generated from the recorded real traffic data published at Arizona Maricopa County website. Based on the generated traffic data, VIPLE programs are developed to implement the traffic simulation based on dynamic changing traffic data.
ContributorsZhang, Zhemin (Author) / Chen, Yinong (Thesis advisor) / Wang, Yalin (Thesis advisor) / De Luca, Gennaro (Committee member) / Arizona State University (Publisher)
Created2022
Description
Little is known about how cognitive and brain aging patterns differ in older adults with autism spectrum disorder (ASD). However, recent evidence suggests that individuals with ASD may be at greater risk of pathological aging conditions than their neurotypical (NT) counterparts. A growing body of research indicates that older adults with ASD may experience accelerated cognitive decline and neurodegeneration as they age, although studies are limited by their cross-sectional design in a population with strong age-cohort effects. Studying aging in ASD and identifying biomarkers to predict atypical aging is important because the population of older individuals with ASD is growing. Understanding the unique challenges faced as autistic adults age is necessary to develop treatments to improve quality of life and preserve independence. In this study, a longitudinal design was used to characterize cognitive and brain aging trajectories in ASD as a function of autistic trait severity. Principal components analysis (PCA) was used to derive a cognitive metric that best explains performance variability on tasks measuring memory ability and executive function. The slope of the integrated persistent feature (SIP) was used to quantify functional connectivity; the SIP is a novel, threshold-free graph theory metric which summarizes the speed of information diffusion in the brain. Longitudinal mixed models were using to predict cognitive and brain aging trajectories (measured via the SIP) as a function of autistic trait severity, sex, and their interaction. The sensitivity of the SIP was also compared with traditional graph theory metrics. It was hypothesized that older adults with ASD would experience accelerated cognitive and brain aging and furthermore, age-related changes in brain network topology would predict age-related changes in cognitive performance.
For both cognitive and brain aging, autistic traits and sex interacted to predict trajectories, such that older men with high autistic traits were most at risk for poorer outcomes. In men with autism, variability in SIP scores across time points trended toward predicting cognitive aging trajectories. Findings also suggested that autistic traits are more sensitive to differences in brain aging than diagnostic group and that the SIP is more sensitive to brain aging trajectories than other graph theory metrics. However, further research is required to determine how physiological biomarkers such as the SIP are associated with cognitive outcomes.
ContributorsSullivan, Georgia (Author) / Braden, Blair (Thesis advisor) / Kodibagkar, Vikram (Thesis advisor) / Schaefer, Sydney (Committee member) / Wang, Yalin (Committee member) / Arizona State University (Publisher)
Created2022
Description
ABSTRACTWith the National Aeronautics and Space Administration (NASA) Psyche Mission, humans will soon have the first opportunity to explore a new kind of planetary body: one composed mostly of metal as opposed to stony minerals or ices. Identifying the composition of asteroids from Earth-based observations has been an ongoing challenge. Although optical reflectance spectra, radar, and orbital dynamics can constrain an asteroid’s mineralogy and bulk density, in many cases there is not a clear or precise match with analogous materials such as meteorites. Additionally, the surfaces of asteroids and other small, airless planetary bodies can be heavily modified over geologic time by exposure to the space environment. To accurately interpret remote sensing observations of metal-rich asteroids, it is therefore necessary to understand how the processes active on asteroid surfaces affect metallic materials. This dissertation represents a first step toward that understanding. In collaboration with many colleagues, I have performed laboratory experiments on iron meteorites to simulate solar wind ion irradiation, surface heating, micrometeoroid bombardment, and high-velocity impacts. Characterizing the meteorite surface’s physical and chemical properties before and after each experiment can constrain the effects of each process on a metal-rich surface in space. While additional work will be needed for a complete understanding, it is nevertheless possible to make some early predictions of what (16) Psyche’s surface regolith might look like when humans observe it up close. Moreover, the results of these experiments will inform future exploration beyond asteroid Psyche as humans attempt to understand how Earth’s celestial neighborhood came to be.
ContributorsChristoph, John Morgan M. (Author) / Elkins-Tanton, Linda (Thesis advisor) / Williams, David (Committee member) / Dukes, Catherine (Committee member) / Sharp, Thomas (Committee member) / Bell III, James (Committee member) / Arizona State University (Publisher)
Created2023
Description
As robots become increasingly integrated into the environments, they need to learn how to interact with the objects around them. Many of these objects are articulated with multiple degrees of freedom (DoF). Multi-DoF objects have complex joints that require specific manipulation orders, but existing methods only consider objects with a single joint. To capture the joint structure and manipulation sequence of any object, I introduce the "Object Kinematic State Machines" (OKSMs), a novel representation that models the kinematic constraints and manipulation sequences of multi-DoF objects. I also present Pokenet, a deep neural network architecture that estimates the OKSMs from the sequence of point cloud data of human demonstrations. I conduct experiments on both simulated and real-world datasets to validate my approach. First, I evaluate the modeling of multi-DoF objects on a simulated dataset, comparing against the current state-of-the-art method. I then assess Pokenet's real-world usability on a dataset collected in my lab, comprising 5,500 data points across 4 objects. Results showcase that my method can successfully estimate joint parameters of novel multi-DoF objects with over 25% more accuracy on average than prior methods.
ContributorsGUPTA, ANMOL (Author) / Gopalan, Nakul (Thesis advisor) / Zhang, Yu (Committee member) / Wang, Yalin (Committee member) / Arizona State University (Publisher)
Created2024
Description
In today's data-driven world, privacy is a significant concern. It is crucial to preserve the privacy of sensitive information while visualizing data. This thesis aims to develop new techniques and software tools that support Vega-Lite visualizations while maintaining privacy. Vega-Lite is a visualization grammar based on Wilkinson's grammar of graphics. The project extends Vega-Lite to incorporate privacy algorithms such as k-anonymity, l-diversity, t-closeness, and differential privacy. This is done by using a unique multi-input loop module logic that generates combinations of attributes as a new anonymization method. Differential privacy is implemented by adding controlled noise (Laplace or Exponential) to the sensitive columns in the dataset. The user defines custom rules in the JSON schema, mentioning the privacy methods and the sensitive column. The schema is validated using Another JSON Validation library, and these rules help identify the anonymization techniques to be performed on the dataset before sending it back to the Vega-Lite visualization server. Multiple datasets satisfying the privacy requirements are generated, and their utility scores are provided so that the user can trade-off between privacy and utility on the datasets based on their requirements. The interface developed is user-friendly and intuitive and guides users in using it. It provides appropriate feedback on the privacy-preserving visualizations generated through various utility metrics. This application is helpful for technical or domain experts across multiple domains where privacy is a big concern, such as medical institutions, traffic and urban planning, financial institutions, educational records, and employer-employee relations. This project is novel as it provides a one-stop solution for privacy-preserving visualization. It works on open-source software, Vega-Lite, which several organizations and users use for business and educational purposes.
ContributorsSekar, Manimozhi (Author) / Bryan, Chris (Thesis advisor) / Wang, Yalin (Committee member) / Cao, Zhichao (Committee member) / Arizona State University (Publisher)
Created2024