Physical Sciences and Mathematics Commons^™

Spatiotemporal Subspace Feature Tracking By Mining Discriminatory Characteristics, Richard D. Appiah

Doctoral Dissertations

Recent advancements in data collection technologies have made it possible to collect heterogeneous data at complex levels of abstraction, and at an alarming pace and volume. Data mining, and most recently data science seek to discover hidden patterns and insights from these data by employing a variety of knowledge discovery techniques. At the core of these techniques is the selection and use of features, variables or properties upon which the data were acquired to facilitate effective data modeling. Selecting relevant features in data modeling is critical to ensure an overall model accuracy and optimal predictive performance of future effects. The …

Motion-Capture-Based Hand Gesture Recognition For Computing And Control, Andrew Gardner

Doctoral Dissertations

This dissertation focuses on the study and development of algorithms that enable the analysis and recognition of hand gestures in a motion capture environment. Central to this work is the study of unlabeled point sets in a more abstract sense. Evaluations of proposed methods focus on examining their generalization to users not encountered during system training.

In an initial exploratory study, we compare various classification algorithms based upon multiple interpretations and feature transformations of point sets, including those based upon aggregate features (e.g. mean) and a pseudo-rasterization of the capture space. We find aggregate feature classifiers to be balanced across …

Dpweka: Achieving Differential Privacy In Weka, Srinidhi Katla

Graduate Theses and Dissertations

Organizations belonging to the government, commercial, and non-profit industries collect and store large amounts of sensitive data, which include medical, financial, and personal information. They use data mining methods to formulate business strategies that yield high long-term and short-term financial benefits. While analyzing such data, the private information of the individuals present in the data must be protected for moral and legal reasons. Current practices such as redacting sensitive attributes, releasing only the aggregate values, and query auditing do not provide sufficient protection against an adversary armed with auxiliary information. In the presence of additional background information, the privacy protection …

Combinatorial Algorithms For Perturbation Theory And Application On Quantum Computing, Yudong Cao

Open Access Dissertations

Quantum computing is an emerging area between computer science and physics. Numerous problems in quantum computing involve quantum many-body interactions. This dissertation concerns the problem of simulating arbitrary quantum many-body interactions using realistic two-body interactions. To address this issue, a general class of techniques called perturbative reductions (or perturbative gadgets) is adopted from quantum complexity theory and in this dissertation these techniques are improved for experimental considerations. The idea of perturbative reduction is based on the mathematical machinery of perturbation theory in quantum physics. A central theme of this dissertation is then to analyze the combinatorial structure of the perturbation …

A Framework For The Statistical Analysis Of Mass Spectrometry Imaging Experiments, Kyle Bemis

Open Access Dissertations

Mass spectrometry (MS) imaging is a powerful investigation technique for a wide range of biological applications such as molecular histology of tissue, whole body sections, and bacterial films , and biomedical applications such as cancer diagnosis. MS imaging visualizes the spatial distribution of molecular ions in a sample by repeatedly collecting mass spectra across its surface, resulting in complex, high-dimensional imaging datasets. Two of the primary goals of statistical analysis of MS imaging experiments are classification (for supervised experiments), i.e. assigning pixels to pre-defined classes based on their spectral profiles, and segmentation (for unsupervised experiments), i.e. assigning pixels to newly …

Computational Environment For Modeling And Analysing Network Traffic Behaviour Using The Divide And Recombine Framework, Ashrith Barthur

Open Access Dissertations

There are two essential goals of this research. The first goal is to design and construct a computational environment that is used for studying large and complex datasets in the cybersecurity domain. The second goal is to analyse the Spamhaus blacklist query dataset which includes uncovering the properties of blacklisted hosts and understanding the nature of blacklisted hosts over time.

The analytical environment enables deep analysis of very large and complex datasets by exploiting the divide and recombine framework. The capability to analyse data in depth enables one to go beyond just summary statistics in research. This deep analysis is …

Divide And Recombined For Large Complex Data: Nonparametric-Regression Modelling Of Spatial And Seasonal-Temporal Time Series, Xiaosu Tong

Open Access Dissertations

In the first chapter of this dissertation, I briefly introduce one type of nonparametric regression method, namely local polynomial regression, followed by emphasis on one specific application of loess on time series decomposition, called Seasonal Trend Loess (STL). The chapter is closed by the introduction of D\&R; (Divide and Recombined) statistical framework. Data can be divided into subsets, each of which is applied with a statistical analysis method. This is an embarrassing parallel procedure since there is no communication between each subset. Then the analysis result for each subset are combined together to be the final analysis outcome for the …

Controlling For Confounding Network Properties In Hypothesis Testing And Anomaly Detection, Timothy La Fond

Open Access Dissertations

An important task in network analysis is the detection of anomalous events in a network time series. These events could merely be times of interest in the network timeline or they could be examples of malicious activity or network malfunction. Hypothesis testing using network statistics to summarize the behavior of the network provides a robust framework for the anomaly detection decision process. Unfortunately, choosing network statistics that are dependent on confounding factors like the total number of nodes or edges can lead to incorrect conclusions (e.g., false positives and false negatives). In this dissertation we describe the challenges that face …

User-Centric Workload Analytics: Towards Better Cluster Management, Suhas Raveesh Javagal

Open Access Theses

Effective management of computing clusters and providing a high quality customer support is not a trivial task. Due to rise of community clusters there is an increase in the diversity of workloads and the user demographic. Owing to this and privacy concerns of the user, it is difficult to identify performance issues, reduce resource wastage and understand implicit user demands. In this thesis, we perform in-depth analysis of user behavior, performance issues, resource usage patterns and failures in the workloads collected from a university-wide community cluster and two clusters maintained by a government lab. We also introduce a set of …

Implementation And Validation Of A Probabilistic Open Source Baseball Engine (Posbe): Modeling Hitters And Pitchers, Rhett Tracy Schaefer

Open Access Theses

This manuscript details the implementation and validation of an open source probabilistic baseball engine (POSBE) that focuses on the hitter and pitcher model of the simulation. The simulation produced outcomes that parallel those observed in actual professional Major League Baseball games. The observed data were taken from the nineteen games played between the New York Yankees (NYY) and Boston Red Sox (BOS) during the 2015 season. The potential hitter/pitcher outcomes of interest were singles, doubles, triples, homeruns, walks, hit-by-pitch, and strikeouts. The nineteen game series was simulated 1000 times, resulting in a total of 19,000 simulations. The eighteen hitters and …

Sensitivity Of Mixed Models To Computational Algorithms Of Time Series Data, Gunaime Nevine

Doctoral Dissertations

Statistical analysis is influenced by implementation of the algorithms used to execute the computations associated with various statistical techniques. Over many years; very important criteria for model comparison has been studied and examined, and two algorithms on a single dataset have been performed numerous times. The goal of this research is not comparing two or more models on one dataset, but comparing models with numerical algorithms that have been used to solve them on the same dataset.

In this research, different models have been broadly applied in modeling and their contrasting which are affected by the numerical algorithms in different …

Performance Modeling And Optimization Techniques For Heterogeneous Computing, Supada Laosooksathit

Doctoral Dissertations

Since Graphics Processing Units (CPUs) have increasingly gained popularity amoung non-graphic and computational applications, known as General-Purpose computation on GPU (GPGPU), CPUs have been deployed in many clusters, including the world's fastest supercomputer. However, to make the most efficiency from a GPU system, one should consider both performance and reliability of the system.

This dissertation makes four major contributions. First, the two-level checkpoint/restart protocol that aims to reduce the checkpoint and recovery costs with a latency hiding strategy in a system between a CPU (Central Processing Unit) and a GPU is proposed. The experimental results and analysis reveals some benefits, …

Methods For Increasing Domains Of Convergence In Iterative Linear System Solvers, David Michael Imberti

Open Access Dissertations

In this thesis, we introduce and improve various methods for increasing the domains of convergence for iterative linear system solvers. We rely on the following three approaches: making the iteration adaptive, or nesting an inner iteration inside of a previously determined outer iteration; using deflation and projections to manipulate the spectra inherent to the iteration; and/or focusing on reordering schemes. We will analyze a specific combination of these three strategies. In particular, we propose to examine the influence of nesting a Flexible Generalized Minimum Residual algorithm together with an inner Recursive Projection Method using a banded preconditioner resulting from the …

Developing A B -Tagging Algorithm Using Soft Muons At Level-3 For The Dø Detector At Fermilab, Mayukh Das

Doctoral Dissertations

The current data-taking phase of the DØ detector at Fermilab, called Run II, is designed to aid the search for the Higgs Boson. The neutral Higgs is postulated to have a mass of 117 GeV. One of the channels promising the presence of this hypothetical particle is through the decay of b-quark into a muon. The process of identifying a b-quark in a jet using muon as a reference is b-tagging with a muon tag.

At the current data taking and analysis rate, it will take long to reach the process of identifying valid events. The triggering mechanism of the …

Integrated Modeling And Parallel Computation Of Laser-Induced Axisymmetric Rod Growth, Hong Lan

Doctoral Dissertations

To fully investigate a pyrolytic Laser-induced chemical vapor deposition (LCVD) system for growing an axisymmetric rod, a novel integrated three-dimensional mathematical model was developed not only to describe the heat transport in the deposit and substrate, but also to simulate the gas-phase in the heated reaction zone and its effect on growth rate. The integrated model consists of three components: the substrate, rod, and gas-phase domains. Each component is a separate model and the three components are dynamically integrated into one model for simulating the iterative and complex process of rod deposition.

The gas-phase reaction is modeled by the gas-phase …

Computational Approaches To The Design And Analysis Of Stability Of Polypeptide Multilayer Thin Films, Bin Zheng

Doctoral Dissertations

The focus of this research is the development of computational approaches to understanding the physical basis of layer-by-layer assembly (LBL), a key methodology of nanomanufacturing. The results provided detailed information on structure which cannot be obtained directly by experiments.

The model systems chosen for study are polypeptide chains. Reasons for this are that polypeptides are no less polyelectrolytes than the more usual polyions, and one can control the primary structure of a polypeptide on a residue-by-residue basis using modern synthetic methods. Moreover, as peptides constitute one of the four major classes of biological macromolecules, research in this direction is expected …

Modeling Of The Inverse Heat -Conduction Problem With Application To Laser Chemical Vapor Deposition And Bioheat Transfer, Peng Zhen

Doctoral Dissertations

This dissertation consists of two parts. Part one deals with three-dimensional laser induced chemical vapor deposition (3D-LCVD), whereas part two deals with a Pennes model of a 3D skin structure. LCVD is an important technique in manufacturing complex micro-structures with high aspect ratio. In part one, a numerical model was developed for simulating kinetically-limited growth of an axisymmetric cylindrical rod by pre-specifying the surface temperature distribution required for growing the rod and then by obtaining optimized laser power that gives rise to the pre-specified temperature distribution. The temperature distribution at the surface of the rod was assumed to be at …

Machine Learning Approaches For Determining Effective Seeds For K -Means Algorithm, Kaveephong Lertwachara

Doctoral Dissertations

In this study, I investigate and conduct an experiment on two-stage clustering procedures, hybrid models in simulated environments where conditions such as collinearity problems and cluster structures are controlled, and in real-life problems where conditions are not controlled. The first hybrid model (NK) is an integration between a neural network (NN) and the k-means algorithm (KM) where NN screens seeds and passes them to KM. The second hybrid (GK) uses a genetic algorithm (GA) instead of the neural network. Both NN and GA used in this study are in their simplest-possible forms.

In the simulated data sets, I investigate two …

Fuzzy Product -Limit Estimators: Soft Computing In The Presence Of Very Small And Highly Censored Data Sets, Kian Lawrence Pokorny

Doctoral Dissertations

When very few data are available and a high proportion of the data is censored, accurate estimates of reliability are problematic. Standard statistical methods require a more complete data set, and with any fewer data, expert knowledge or heuristic methods are required. In the current research a computational system is developed that obtains a survival curve, point estimate, and confidence interval about the point estimate.

The system uses numerical methods to define fuzzy membership functions about each data point that quantify uncertainty due to censoring. The “fuzzy” data are then used to estimate a survival curve, and the mean survival …

A Hybrid Finite Element-Finite Difference Method For Thermal Analysis In A Double-Layered Thin Film, Teng Zhu

Doctoral Dissertations

Thin film technology is of vital importance in microtechnology applications. For instance, thin films of metals, of dielectrics such as SiO2, or Si semiconductors are important components of microelectronic devices. The reduction of the device size to the microscale has the advantage of enhancing the switching speed of the device. The reduction, on the other hand, increases the rate of heat generation that leads to a high thermal load on the microdevice. Heat transfer at the microscale with an ultrafast pulsed-laser is also a very important process for thin films. Hence, studying the thermal behavior of thin films or of …