Search Content

Toward Reliable Graph Matching: from Deterministic Optimization to Combinatorial Learning

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…

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

Toward Reliable Graph Matching: from Deterministic Optimization to Combinatorial Learning

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…

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

Cognitive Aging and Flexibility in Canids

Description

This dissertation describes a series of four studies on cognitive aging, working memory, and cognitive flexibility in dogs (Canis lupus familiaris) and their wild relatives. In Chapters 2 and 3, I designed assessments for age-related cognitive deficits in pet dogs which can be deployed rapidly using inexpensive and accessible materials.…

This dissertation describes a series of four studies on cognitive aging, working memory, and cognitive flexibility in dogs (Canis lupus familiaris) and their wild relatives. In Chapters 2 and 3, I designed assessments for age-related cognitive deficits in pet dogs which can be deployed rapidly using inexpensive and accessible materials. These novel tests can be easily implemented by owners, veterinarians, and clinicians and therefore, may improve care for elderly dogs by aiding in the diagnosis of dementia. In addition, these widely deployable tests may facilitate the use of dementia in pet dogs as a naturally occurring model of Alzheimer’s Disease in humans.In Chapters 4 and 5, I modified one of these tests to demonstrate for the first time that coyotes (Canis latrans) and wolves (Canis lupus lupus) develop age-related deficits in cognitive flexibility. This was an important first step towards differentiating between the genetic and environmental components of dementia in dogs and in turn, humans. Unexpectedly, I also detected cognitive deficits in young, adult dogs and wolves but not coyotes. These finding add to a recent shift in understanding cognitive development in dogs which may improve cognitive aging tests as well as training, care, and use of working and pet dogs. These findings also suggest that the ecology of coyotes may select for flexibility earlier in development. In Chapter 5, I piloted the use of the same cognitive flexibility test for red and gray foxes so that future studies may test for lifespan changes in the cognition of small-bodied captive canids. More broadly, this paradigm may accommodate physical and behavioral differences between diverse pet and captive animals. In Chapters 4 and 5, I examined which ecological traits drive the evolution of behavioral flexibility and in turn, species resilience. I found that wolves displayed less flexibility than dogs and coyotes suggesting that species which do not rely heavily on unstable resources may be ill-equipped to cope with human habitat modification. Ultimately, this comparative work may help conservation practitioners to identify and protect species that cannot cope with rapid and unnatural environmental change.

ContributorsVan Bourg, Joshua (Author) / Wynne, Clive D (Thesis advisor) / Aktipis, C. Athena (Committee member) / Gilby, Ian C (Committee member) / Young, Julie K (Committee member) / Arizona State University (Publisher)

Created2022

A Continuous Learning Approach to Alzheimer’s Disease Progression Modeling and Domain Adaptation among Imaging Cohorts

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…

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

Reducing Single-Use Plastic & Nitrile Waste in U.S. Research Institutions

Description

The production and incineration of single-use micropipette tips and disposable gloves, which are heavily used within laboratory facilities, generate large amounts of greenhouse gasses (GHGs) and accelerate climate change. Plastic waste that is not incinerated often is lost in the environment. The long degradation times associated with this waste exacerbates…

The production and incineration of single-use micropipette tips and disposable gloves, which are heavily used within laboratory facilities, generate large amounts of greenhouse gasses (GHGs) and accelerate climate change. Plastic waste that is not incinerated often is lost in the environment. The long degradation times associated with this waste exacerbates a variety of environmental problems such as substance runoff and ocean pollution. The objective of this study was to evaluate the efficacy of possible solutions for minimizing micropipette tip and disposable glove waste within laboratory spaces. It was hypothesized that simultaneously implementing the use of micropipette tip washers (MTWs) and energy-from-glove-waste programs (EGWs) would significantly reduce (p < 0.05) the average combined annual single-use plastic micropipette tip and nitrile glove waste (in kg) per square meter of laboratory space in the United States. ASU’s Biodesign Institute (BDI) was used as a case study to inform on the thousands of different laboratory facilities that exist all across the United States. Four separate research laboratories within the largest public university of the U.S. were sampled to assess the volume of plastic waste from single-use micropipette tips and gloves. Resultant data were used to represent the totality of single-use waste from the case study location and then extrapolated to all laboratory space in the United States. With the implementation of EGWs, annual BDI glove waste is reduced by 100% (0.47 ± 0.26 kg/m2; 35.5 ± 19.3 metric tons total) and annual BDI glove-related carbon emissions are reduced by ~5.01% (0.165 ± 0.09 kg/m2; 1.24 ± 0.68 metric tons total). With the implementation of MTWs, annual BDI micropipette tip waste is reduced by 92% (0.117 ± 0.03 kg/m2; 0.88 ± 0.25 metric tons total) and annual BDI tip-related carbon emissions are reduced by ~83.6% (4.04 ± 1.25 kg/m2; 30.5 ± 9.43 metric tons total). There was no significant difference (p = 0.06) observed between the mass of single-use waste (kg) in the sampled laboratory spaces before (x̄ = 47.1; σ = 43.3) and after (x̄ =0.070; σ = 0.033) the implementation of the solutions. When examining both solutions (MTWs & EGWs) implemented in conjunction with one another, the annual BDI financial savings (in regard to both purchasing and disposal costs) after the first year were determined to be ~$7.92 ± $9.31/m2 (7,500 m2 of total wet laboratory space) or ~$60,000 ± $70,000 total. These savings represent ~15.77% of annual BDI spending on micropipette tips and nitrile gloves. The large error margins in these financial estimates create high uncertainty for whether or not BDI would see net savings from implementing both solutions simultaneously. However, when examining the implementation of only MTWs, the annual BDI financial savings (in regard to both purchasing and disposal costs) after the first year were determined to be ~$12.01 ± $6.79 kg/m2 or ~$91,000 ± $51,200 total. These savings represent ~23.92% of annual BDI spending on micropipette tips and nitrile gloves. The lower error margins for this estimate create a much higher likelihood of net savings for BDI. Extrapolating to all laboratory space in the United States, the total annual amount of plastic waste avoided with the implementation of the MTWs was identified as 8,130 ± 2,290 tons or 0.023% of all solid plastic waste produced in the United States in 2018. The total amount of nitrile waste avoided with the implementation of the EGWs was identified as 32,800 ± 17,900 tons or 0.36% of all rubber solid waste produced in the United States in 2018. The total amount of carbon emissions avoided with the implementation of the MTWs was identified as 281,000 ± 87,000 tons CO2eq or 5.4*10-4 % of all CO2eq GHG emissions produced in the United States in 2020. Both the micropipette tip washer and the glove waste avoidance program solutions can be easily integrated into existing laboratories without compromising the integrity of the activities taking place. Implemented on larger scales, these solutions hold the potential for significant single-use waste reduction.

ContributorsZdrale, Gabriel (Author) / Mahant, Akhil (Co-author) / Halden, Rolf (Thesis director) / Biyani, Nivedita (Committee member) / Driver, Erin (Committee member) / Barrett, The Honors College (Contributor) / Harrington Bioengineering Program (Contributor)

Created2022-05

Population Receptive Field Prediction with Convolutional Neural Networks

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…

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

Reducing Single-Use Plastic & Nitrile Waste in U.S. Research Institutions

Description

The production and incineration of single-use micropipette tips and disposable gloves, which are heavily used within laboratory facilities, generate large amounts of greenhouse gasses (GHGs) and accelerate climate change. Plastic waste that is not incinerated often is lost in the environment. The long degradation times associated with this waste exacerbates…

The production and incineration of single-use micropipette tips and disposable gloves, which are heavily used within laboratory facilities, generate large amounts of greenhouse gasses (GHGs) and accelerate climate change. Plastic waste that is not incinerated often is lost in the environment. The long degradation times associated with this waste exacerbates a variety of environmental problems such as substance runoff and ocean pollution. The objective of this study was to evaluate the efficacy of possible solutions for minimizing micropipette tip and disposable glove waste within laboratory spaces. It was hypothesized that simultaneously implementing the use of micropipette tip washers (MTWs) and energy-from-glove-waste programs (EGWs) would significantly reduce (p < 0.05) the average combined annual single-use plastic micropipette tip and nitrile glove waste (in kg) per square meter of laboratory space in the United States. ASU’s Biodesign Institute (BDI) was used as a case study to inform on the thousands of different laboratory facilities that exist all across the United States. Four separate research laboratories within the largest public university of the U.S. were sampled to assess the volume of plastic waste from single-use micropipette tips and gloves. Resultant data were used to represent the totality of single-use waste from the case study location and then extrapolated to all laboratory space in the United States. With the implementation of EGWs, annual BDI glove waste is reduced by 100% (0.47 ± 0.26 kg/m2; 35.5 ± 19.3 metric tons total) and annual BDI glove-related carbon emissions are reduced by ~5.01% (0.165 ± 0.09 kg/m2; 1.24 ± 0.68 metric tons total). With the implementation of MTWs, annual BDI micropipette tip waste is reduced by 92% (0.117 ± 0.03 kg/m2; 0.88 ± 0.25 metric tons total) and annual BDI tip-related carbon emissions are reduced by ~83.6% (4.04 ± 1.25 kg/m2; 30.5 ± 9.43 metric tons total). There was no significant difference (p = 0.06) observed between the mass of single-use waste (kg) in the sampled laboratory spaces before (x̄ = 47.1; σ = 43.3) and after (x̄ =0.070; σ = 0.033) the implementation of the solutions.When examining both solutions (MTWs & EGWs) implemented in conjunction with one another, the annual BDI financial savings (in regard to both purchasing and disposal costs) after the first year were determined to be ~$7.92 ± $9.31/m2 (7,500 m2 of total wet laboratory space) or ~$60,000 ± $70,000 total. These savings represent ~15.77% of annual BDI spending on micropipette tips and nitrile gloves. The large error margins in these financial estimates create high uncertainty for whether or not BDI would see net savings from implementing both solutions simultaneously. However, when examining the implementation of only MTWs, the annual BDI financial savings (in regard to both purchasing and disposal costs) after the first year were determined to be ~$12.01 ± $6.79 kg/m2 or ~$91,000 ± $51,200 total. These savings represent ~23.92% of annual BDI spending on micropipette tips and nitrile gloves. The lower error margins for this estimate create a much higher likelihood of net savings for BDI. Extrapolating to all laboratory space in the United States, the total annual amount of plastic waste avoided with the implementation of the MTWs was identified as 8,130 ± 2,290 tons or 0.023% of all solid plastic waste produced in the United States in 2018. The total amount of nitrile waste avoided with the implementation of the EGWs was identified as 32,800 ± 17,900 tons or 0.36% of all rubber solid waste produced in the United States in 2018. The total amount of carbon emissions avoided with the implementation of the MTWs was identified as 281,000 ± 87,000 tons CO2eq or 5.4*10-4 % of all CO2eq GHG emissions produced in the United States in 2020. Both the micropipette tip washer and the glove waste avoidance program solutions can be easily integrated into existing laboratories without compromising the integrity of the activities taking place. Implemented on larger scales, these solutions hold the potential for significant single-use waste reduction.

ContributorsMahant, Akhil (Author) / Zdrale, Gabriel (Co-author) / Halden, Rolf (Thesis director) / Biyani, Nivedita (Committee member) / Driver, Erin (Committee member) / Barrett, The Honors College (Contributor) / School of International Letters and Cultures (Contributor) / School of Life Sciences (Contributor)

Created2022-05

Assessing the Potential to Determine Diabetes Prevalence at the Community-Level Using Wastewater-Based Epidemiology

Description

Diabetes affects millions of people globally and can lead to other severe health complications when undiagnosed or not properly managed. The incidence of diabetes has rapidly increased over the past several years, however, not all individuals have access to affordable or convenient healthcare. We hypothesize that wastewater-based epidemiology (WBE) has…

Diabetes affects millions of people globally and can lead to other severe health complications when undiagnosed or not properly managed. The incidence of diabetes has rapidly increased over the past several years, however, not all individuals have access to affordable or convenient healthcare. We hypothesize that wastewater-based epidemiology (WBE) has the potential to assess community health status by analyzing biomarkers indicative of human health and disease, including diabetes. Used in tandem with current methods, monitoring indicators of diabetes in community wastewater could provide a comprehensive assessment tool for disease prevalence in large and small populations. Specifically, the proposed targeted biomarker evaluated in this study to indicate population-wide diabetes prevalence was 8-hydroxy-2’- deoxyguanosine (8-OHdG). This work combines a rigorous literature review and initial laboratory studies to explore the possibility of diabetes monitoring at the community level using WBE. Here, 24-hour composite wastewater samples were collected from within two wastewater sub-catchments of Greater Tempe, AZ. Overall goals of this study were to: i) Determine the feasibility to detect endogenous markers of diabetes in community wastewater; ii) Assess the potential impact of confounding factors, such as smoking, cancer, and atherosclerosis, through a literature analysis; and iii) Evaluate the socioeconomic status and demographics of the study population. Preliminary results of the experiments suggest this methodology to be feasible, as indicated by the observation of detectable signals of 8-OHdG in community wastewater collected from the sewer infrastructure; however, future work and continued experimentation will be required to address low signal intensity and assay precision and accuracy. Thus, the work presented here provides valuable proof-of-concept data, with detailed information on the method employed and identified opportunities to further determine the relationship between 8-OHdG concentrations in municipal wastewater and diabetes prevalence at the community level.

ContributorsNguyen, Jasmine (Author) / John, Dona (Co-author) / Halden, Rolf (Thesis director) / Driver, Erin (Committee member) / Bowes, Devin (Committee member) / Barrett, The Honors College (Contributor) / School of Molecular Sciences (Contributor) / Dean, W.P. Carey School of Business (Contributor) / Department of Finance (Contributor)

Created2022-05

Assessing the Potential to Determine Diabetes Prevalence at the Community-Level Using Wastewater-Based Epidemiology

Description

Diabetes affects millions of people globally and can lead to other severe health complications when undiagnosed or not properly managed. The incidence of diabetes has rapidly increased over the past several years, however, not all individuals have access to affordable or convenient healthcare. We hypothesize that wastewater-based epidemiology (WBE) has…

Diabetes affects millions of people globally and can lead to other severe health complications when undiagnosed or not properly managed. The incidence of diabetes has rapidly increased over the past several years, however, not all individuals have access to affordable or convenient healthcare. We hypothesize that wastewater-based epidemiology (WBE) has the potential to assess community health status by analyzing biomarkers indicative of human health and disease, including diabetes. Used in tandem with current methods, monitoring indicators of diabetes in community wastewater could provide a comprehensive assessment tool for disease prevalence in large and small populations. Specifically, the proposed targeted biomarker evaluated in this study to indicate population-wide diabetes prevalence was 8-hydroxy-2’-deoxyguanosine (8-OHdG). This work combines a rigorous literature review and initial laboratory studies to explore the possibility of diabetes monitoring at the community level using WBE. Here, 24-hour composite wastewater samples were collected from within two wastewater sub-catchments of Greater Tempe, AZ. Overall goals of this study were to: i) Determine the feasibility to detect endogenous markers of diabetes in community wastewater; ii) Assess the potential impact of confounding factors, such as smoking, cancer, and atherosclerosis, through a literature analysis; and iii) Evaluate the socioeconomic status and demographics of the study population. Preliminary results of the experiments suggest this methodology to be feasible, as indicated by the observation of detectable signals of 8-OHdG in community wastewater collected from the sewer infrastructure; however, future work and continued experimentation will be required to address low signal intensity and assay precision and accuracy. Thus, the work presented here provides valuable proof-of-concept data, with detailed information on the method employed and identified opportunities to further determine the relationship between 8-OHdG concentrations in municipal wastewater and diabetes prevalence at the community level.

ContributorsJohn, Dona (Author) / Nguyen, Jasmine (Co-author) / Halden, Rolf (Thesis director) / Driver, Erin (Committee member) / Bowes, Devin (Committee member) / Barrett, The Honors College (Contributor) / School of Molecular Sciences (Contributor) / Department of Psychology (Contributor)

Created2022-05

Correlation Between Regional Toxin Exposures and Risk of Parkinson's Disease: A Meta-Analysis

Description

Currently, there are a number of studies confirming the link between exposure to certain chemicals, notably pesticides (Costello et. al. 2009, Wang et. al 2014), heavy metals such as arsenic (Chen et. al. 2017), ambient air pollution (Chen et. al. 2016), and chemicals specific to certain industrial fields (Nielsen et.…

Currently, there are a number of studies confirming the link between exposure to certain chemicals, notably pesticides (Costello et. al. 2009, Wang et. al 2014), heavy metals such as arsenic (Chen et. al. 2017), ambient air pollution (Chen et. al. 2016), and chemicals specific to certain industrial fields (Nielsen et. al. 2021). However, few papers have attempted to perform a widespread analysis of the factors associated with Parkinson’s disease to identify whether the risk of developing the disease is dependent on different factors regionally. The goal of my thesis project is to complete a meta-analysis of toxins- where exposure may occur in both residential and occupational settings- that are associated with Parkinson’s to determine such regional differences and to identify any gaps in current literature, which may direct the course of future research in the field. As seen in this paper, it appears that occupational exposure to toxins appears to have the greatest impact on the risk of developing Parkinson’s disease, particularly pesticides and industrial toxins. However, there are numerous gaps with regards to data collection, regions studied, and quantification of toxin concentrations. However, this data may be useful in identifying at-risk populations if more extensive incremental and biopsy data regarding these toxins is provided.

ContributorsAravindan, Anumitha (Author) / Halden, Rolf (Thesis director) / Driver, Erin (Committee member) / Newell, Melanie (Committee member) / Barrett, The Honors College (Contributor) / School of Life Sciences (Contributor)

Created2022-05