Computer Engineering Commons^™

Feature Space Modeling For Accurate And Efficient Learning From Non-Stationary Data, Ayesha Akter

Doctoral Dissertations

A non-stationary dataset is one whose statistical properties such as the mean, variance, correlation, probability distribution, etc. change over a specific interval of time. On the contrary, a stationary dataset is one whose statistical properties remain constant over time. Apart from the volatile statistical properties, non-stationary data poses other challenges such as time and memory management due to the limitation of computational resources mostly caused by the recent advancements in data collection technologies which generate a variety of data at an alarming pace and volume. Additionally, when the collected data is complex, managing data complexity, emerging from its dimensionality and …

Multi-Segment Multi-Criteria Approach For Selection Of Trenchless Construction Methods, Ashikul Islam

Doctoral Dissertations

The research work presented in this thesis has two broad objectives as well as five individual goals. The first objective is to search and determine the minimum cost and corresponding goodness-of-fit by using a different combination of methods that are capable of resolving the problem that exists in multiple segments. This approach can account for variations in unit price and the cost of the design and the inspection associated with multiple methods. The second objective is to calculate the minimum risk for the preferred solution set. The five individual goals are 1) reduction in total cost, 2) application of Genetic …

Using Power-Law Properties Of Social Groups For Cloud Defense And Community Detection, Justin L. Rice

Doctoral Dissertations

The power-law distribution can be used to describe various aspects of social group behavior. For mussels, sociobiological research has shown that the Lévy walk best describes their self-organizing movement strategy. A mussel's step length is drawn from a power-law distribution, and its direction is drawn from a uniform distribution. In the area of social networks, theories such as preferential attachment seek to explain why the degree distribution tends to be scale-free. The aim of this dissertation is to glean insight from these works to help solve problems in two domains: cloud computing systems and community detection.

Privacy and security are …

Reliability Models For Hpc Applications And A Cloud Economic Model, Thanadech Thanakornworakij

Doctoral Dissertations

With the enormous number of computing resources in HPC and Cloud systems, failures become a major concern. Therefore, failure behaviors such as reliability, failure rate, and mean time to failure need to be understood to manage such a large system efficiently.

This dissertation makes three major contributions in HPC and Cloud studies. First, a reliability model with correlated failures in a k-node system for HPC applications is studied. This model is extended to improve accuracy by accounting for failure correlation. Marshall-Olkin Multivariate Weibull distribution is improved by excess life, conditional Weibull, to better estimate system reliability. Also, the univariate …

Sub Pixel Analysis And Processing Of Sensor Data For Mobile Target Intelligence Information And Verification, Theresa Allen Williams

Doctoral Dissertations

This dissertation introduces a novel process to study and analyze sensor data in order to obtain information pertaining to mobile targets at the sub-pixel level. The process design is modular in nature and utilizes a set of algorithmic tools for change detection, target extraction and analysis, super-pixel processing and target refinement. The scope of this investigation is confined to a staring sensor that records data of sub-pixel vehicles traveling horizontally across the ground. Statistical models of the targets and background are developed with noise and jitter effects. Threshold Change Detection, Duration Change Detection and Fast Adaptive Power Iteration (FAPI) Detection …

Data Mining Based Learning Algorithms For Semi-Supervised Object Identification And Tracking, Michael P. Dessauer

Doctoral Dissertations

Sensor exploitation (SE) is the crucial step in surveillance applications such as airport security and search and rescue operations. It allows localization and identification of movement in urban settings and can significantly boost knowledge gathering, interpretation and action. Data mining techniques offer the promise of precise and accurate knowledge acquisition techniques in high-dimensional data domains (and diminishing the “curse of dimensionality” prevalent in such datasets), coupled by algorithmic design in feature extraction, discriminative ranking, feature fusion and supervised learning (classification). Consequently, data mining techniques and algorithms can be used to refine and process captured data and to detect, recognize, classify, …

Nonlinear Control Strategy For A Cost Effective Myoelectric Prosthetic Hand, Cristian Federico Pasluosta

Doctoral Dissertations

The loss of a limb tremendously impacts the life of the affected individual. In the past decades, researchers have been developing artificial limbs that may return some of the missing functions and cosmetics. However, the development of dexterous mechanisms capable of mimicking the function of the human hand is a complex venture. Even though myoelectric prostheses have advanced, several issues remain to be solved before an artificial limb may be comparable to its human counterpart. Moreover, the high cost of advanced limbs prevents their widespread use among the low-income population.

This dissertation presents a strategy for the low-level of control …