Computer Sciences Commons^™

Learning Robotic Manipulation From User Demonstrations, Rouhollah Rahmatizadeh

Electronic Theses and Dissertations

Personal robots that help disabled or elderly people in their activities of daily living need to be able to autonomously perform complex manipulation tasks. Traditional approaches to this problem employ task-specific controllers. However, these must to be designed by expert programmers, are focused on a single task, and will perform the task as programmed, not according to the preferences of the user. In this dissertation, we investigate methods that enable an assistive robot to learn to execute tasks as demonstrated by the user. First, we describe a learning from demonstration (LfD) method that learns assistive tasks that need to be …

Super Resolution Of Wavelet-Encoded Images And Videos, Vildan Atalay

Electronic Theses and Dissertations

In this dissertation, we address the multiframe super resolution reconstruction problem for wavelet-encoded images and videos. The goal of multiframe super resolution is to obtain one or more high resolution images by fusing a sequence of degraded or aliased low resolution images of the same scene. Since the low resolution images may be unaligned, a registration step is required before super resolution reconstruction. Therefore, we first explore in-band (i.e. in the wavelet-domain) image registration; then, investigate super resolution. Our motivation for analyzing the image registration and super resolution problems in the wavelet domain is the growing trend in wavelet-encoded imaging, …

Hashing For Multimedia Similarity Modeling And Large-Scale Retrieval, Kai Li

Electronic Theses and Dissertations

In recent years, the amount of multimedia data such as images, texts, and videos have been growing rapidly on the Internet. Motivated by such trends, this thesis is dedicated to exploiting hashing-based solutions to reveal multimedia data correlations and support intra-media and inter-media similarity search among huge volumes of multimedia data. We start by investigating a hashing-based solution for audio-visual similarity modeling and apply it to the audio-visual sound source localization problem. We show that synchronized signals in audio and visual modalities demonstrate similar temporal changing patterns in certain feature spaces. We propose to use a permutation-based random hashing technique …

Weakly Labeled Action Recognition And Detection, Waqas Sultani

Electronic Theses and Dissertations

Research in human action recognition strives to develop increasingly generalized methods that are robust to intra-class variability and inter-class ambiguity. Recent years have seen tremendous strides in improving recognition accuracy on ever larger and complex benchmark datasets, comprising realistic actions "in the wild" videos. Unfortunately, the all-encompassing, dense, global representations that bring about such improvements often benefit from the inherent characteristics, specific to datasets and classes, that do not necessarily reflect knowledge about the entity to be recognized. This results in specific models that perform well within datasets but generalize poorly. Furthermore, training of supervised action recognition and detection methods …

Blockchains For Publicizing Available Scientific Datasets, Shirish Patel

Electronic Theses and Dissertations

This thesis explores the effectiveness of blockchain technology for advertisement of scientific data. Recently the advancement in hardware and software for data processing increases the supply and demand for huge data sets. Such data may be widely distributed, and not immediately available to the scientists who need it. We need a method of advertising available datasets to interested parties. Blockchains are a recent innovation developed by the cryptocurrency community, but are increasingly applied to other problem domains. Due to their currency heritage, however, the properties of blockchains do not always lend themselves to new applications. We have developed a prototype …

A Work-Stealing For Dynamic Workload Balancing On Cpu-Gpu Heterogeneous Computing Platforms, Esraa A. Gad

Electronic Theses and Dissertations

Although many general purpose workloads have been accelerated on graphical processing units (gpus) over the last decade, other applications whose runtime behaviors are dynamic and irregular such as ones based on trees and graphs have suffered from serious workload imbalance problem caused by architectural differences between cpu and gpu processors. In this thesis, we propose a work-stealing framework to overcome such problems. Our proposed framework allows cpu and gpu threads to steal tasks from each other as well as within the same device by leveraging fine-grained data sharing and thread communication feature available on modern cpu-gpu heterogeneous systems. The implementation …

Reasoning About Frame Properties In Object-Oriented Programs, Yuyan Bao

Electronic Theses and Dissertations

Framing is important for specification and verification of object-oriented programs. This dissertation develops the local reasoning approach for framing in the presence of data structures with unrestricted sharing and subtyping. It can verify shared data structures specified in a concise way by unifying fine-grained region logic and separation logic. Then the fine-grained region logic is extended to reason about subtyping. First, fine-grained region logic is adapted from region logic to express regions at the granularity of individual fields. Conditional region expressions are introduced; not only does this allow one to specify more precise frame conditions, it also has the ability …

Genetic Algorithm For University Course Timetabling Problem, Achini Kumari Herath

Electronic Theses and Dissertations

Creating timetables for institutes which deal with transport, sport, workforce, courses, examination schedules, and healthcare scheduling is a complex problem. It is difficult and time consuming to solve due to many constraints. Depending on whether the constraints are essential or desirable they are categorized as ‘hard’ and ‘soft’, respectively. Two types of timetables, namely, course and examination are designed for academic institutes. A feasible course timetable could be described as a plan for the movement of students and staff from one classroom to another, without conflicts. Being an NP-complete problem, many attempts have been made using varying computational methods to …

Jml Template Generation, Kushal Raghav Poojari

Electronic Theses and Dissertations

The Java Modeling Language (JML) is a behavioral interface specific language designed to specify Java modules (which are Java classes and interfaces). Specifications are used to describe the intended functionality without considering the way it is implemented. In JML, if a user wants to write specifications for a Java file, he or she must undertake several steps. To help automate the process of creating annotations for method specifications, a tool Jmlspec was created. Jmlspec generated a file that refines the source file and has empty placeholders in which one can write specifications. Although Jmlspec worked with older versions of Java, …

Visual Saliency Detection And Semantic Segmentation, Nasim Souly

Electronic Theses and Dissertations

Visual saliency is the ability to select the most relevant data in the scene and reduce the amount of data that needs to be processed. We propose a novel unsupervised approach to detect visual saliency in videos. For this, we employ a hierarchical segmentation technique to obtain supervoxels of a video, and simultaneously, we build a dictionary from cuboids of the video. Then we create a feature matrix from coefficients of dictionary elements. Next, we decompose this matrix into sparse and redundant parts and obtain salient regions using group lasso. Our experiments provide promising results in terms of predicting eye …

Smart Image Search System Using Personalized Semantic Search Method, Fangyu Zhang

Electronic Theses and Dissertations

Due to the emerge in huge numbers of information on the internet nowadays, search technologies are widely used in various fields. Achieving the most relevant search result for the users becomes a big challenge now. While the traditional semantic search technologies seem to achieve the most relevant search result, however, it faces two problems: one is the one-size-fits-all problem, and another is low efficiency. The purpose of this research is to build a Smart Image Search System by using the personalized semantic search method to solve those problems. The personalized semantic search method makes the search system avoids the one-size-fits-all …

A Reduced Labeled Samples (Rls) Framework For Classification Of Imbalanced Concept-Drifting Streaming Data., Elaheh Arabmakki

Electronic Theses and Dissertations

Stream processing frameworks are designed to process the streaming data that arrives in time. An example of such data is stream of emails that a user receives every day. Most of the real world data streams are also imbalanced as is in the stream of emails, which contains few spam emails compared to a lot of legitimate emails. The classification of the imbalanced data stream is challenging due to the several reasons: First of all, data streams are huge and they can not be stored in the memory for one time processing. Second, if the data is imbalanced, the accuracy …

An Algorithm For The Machine Calculation Of Minimal Paths, Robert Whitinger

Electronic Theses and Dissertations

Problems involving the minimization of functionals date back to antiquity. The mathematics of the calculus of variations has provided a framework for the analytical solution of a limited class of such problems. This paper describes a numerical approximation technique for obtaining machine solutions to minimal path problems. It is shown that this technique is applicable not only to the common case of finding geodesics on parameterized surfaces in R³, but also to the general case of finding minimal functionals on hypersurfaces in Rⁿ associated with an arbitrary metric.

Sparse Feature Learning For Image Analysis In Segmentation, Classification, And Disease Diagnosis., Ehsan Hosseini-Asl

Electronic Theses and Dissertations

The success of machine learning algorithms generally depends on intermediate data representation, called features that disentangle the hidden factors of variation in data. Moreover, machine learning models are required to be generalized, in order to reduce the specificity or bias toward the training dataset. Unsupervised feature learning is useful in taking advantage of large amount of unlabeled data, which is available to capture these variations. However, learned features are required to capture variational patterns in data space. In this dissertation, unsupervised feature learning with sparsity is investigated for sparse and local feature extraction with application to lung segmentation, interpretable deep …

Implementing Agile Development At Scale: An Industry Case Study, Nikita Kataria

Electronic Theses and Dissertations

Agile software development methodologies are extremely popular. Their dynamic restructuring of the development process has been seen as the silver bullet for increasing the productivity of software development. A significant number of studies have analyzed the impact of implementing agile techniques. However these are mostly evaluated only in smaller team settings. There is very little reporting done on how agile development methods can be implemented at the team level and scaled up at the program/portfolio level in large software organizations.

We present the results of an empirical study conducted at Pearson Education. The study focuses on the penetration of agile …

Optimizing Vehicle Usage Using Csp, Sat And Max-Sat, Raheem T. Al Rammahi

Electronic Theses and Dissertations

Most of the companies in Iraq spend significant amounts of time and money when transferring employees between home and work. In this thesis, we model the problem of the Dhi Qar Oil company (DQOC) transportations using three modeling languages from AI: Constraint Programing (CP), Boolean Satisfiability (SAT), and Maximum Satisfiability (MAX-SAT). We then use solvers to find optimal solutions to this problem.

We show which of these solvers is more efficient when finding optimal solutions. For this purpose, we create a test suite of 360 problems to test these solvers. All solvers are applied to these problems and the final …

Leveraging Client Processing For Location Privacy In Mobile Local Search, Wisam Mohamed Eltarjaman

Electronic Theses and Dissertations

Usage of mobile services is growing rapidly. Most Internet-based services targeted for PC based browsers now have mobile counterparts. These mobile counterparts often are enhanced when they use user's location as one of the inputs. Even some PC-based services such as point of interest Search, Mapping, Airline tickets, and software download mirrors now use user's location in order to enhance their services. Location-based services are exactly these, that take the user's location as an input and enhance the experience based on that. With increased use of these services comes the increased risk to location privacy. The location is considered an …

A Near-To-Far Learning Framework For Terrain Characterization Using An Aerial/Ground-Vehicle Team, Ashkan Hajjam

Electronic Theses and Dissertations

In this thesis, a novel framework for adaptive terrain characterization of untraversed far terrain in a natural outdoor setting is presented. The system learns the association between visual appearance of different terrain and the proprioceptive characteristics of that terrain in a self-supervised framework. The proprioceptive characteristics of the terrain are acquired by inertial sensors recording measurements of one second traversals that are mapped into the frequency domain and later through a clustering technique classified into discrete proprioceptive classes. Later, these labels are used as training inputs to the adaptive visual classifier. The visual classifier uses images captured by an aerial …

Greenc5: An Adaptive, Energy-Aware Collection For Green Software Development, Junya Michanan

Electronic Theses and Dissertations

Dynamic data structures in software applications have been shown to have a large impact on system performance. In this paper, we explore energy saving opportunities of interface-based dynamic data structures. Our results suggest that savings opportunities exist in the C5 Collection between 16.95% and 97.50%. We propose a prototype and architecture for creating adaptive green data structures by applying machine learning tools to build a model for predicting energy efficient data structures based on the dynamic workload. Our neural network model can classify energy efficient data structures based on features such as the number of elements, frequency of operations, interface …

Enhancements To Hierarchical Pathfinding Algorithms, Xin Li

Electronic Theses and Dissertations

In this thesis we study the problem of pathfinding in static grid-based maps. We apply the approach of abstraction and refinement. We abstract the grid map into a graph representation, and use the classic A* algorithm to search for a path in the abstract space, and then refine it into low-level path.

We started with a 2013 entry program to the Grid-based Path Planning Competition, and implemented several enhancements to experiment with the tradeoff between memory usage and search speed. Our program returns the refined low-level path incrementally, therefore reduces the first-move lag in large maps. We cache the low-level …

Model Based Security Testing For Autonomous Vehicles, Seana Lisa Hagerman

Electronic Theses and Dissertations

The purpose of this dissertation is to introduce a novel approach to generate a security test suite to mitigate malicious attacks on an autonomous system. Our method uses model based testing (MBT) methods to model system behavior, attacks and mitigations as independent threads in an execution stream. The threads intersect at a rendezvous or attack point. We build a security test suite from a behavioral model, an attack type and a mitigation model using communicating extended finite state machine (CEFSM) models. We also define an applicability matrix to determine which attacks are possible with which states. Our method then builds …

Peccit: An Omniscient Debugger For Web Development, Zachary Ryan Azar

Electronic Theses and Dissertations

Debugging can be an extremely expensive and time-consuming task for a software developer. To find a bug, the developer typically needs to navigate backwards through infected states and symptoms of the bug to find the initial defect. Modern debugging tools are not designed for navigating back-in-time and typically require the user to jump through hoops by setting breakpoints, re-executing, and guessing where errors occur. Omniscient debuggers offer back-in-time debugging capabilities to make this task easier. These debuggers trace the program allowing the user to navigate forwards and backwards through the execution, examine variable histories, and visualize program data and control …

Spatial And Temporal Modeling For Human Activity Recognition From Multimodal Sequential Data, Jun Ye

Electronic Theses and Dissertations

Human Activity Recognition (HAR) has been an intense research area for more than a decade. Different sensors, ranging from 2D and 3D cameras to accelerometers, gyroscopes, and magnetometers, have been employed to generate multimodal signals to detect various human activities. With the advancement of sensing technology and the popularity of mobile devices, depth cameras and wearable devices, such as Microsoft Kinect and smart wristbands, open a unprecedented opportunity to solve the challenging HAR problem by learning expressive representations from the multimodal signals recording huge amounts of daily activities which comprise a rich set of categories. Although competitive performance has been …

Breast Cancer Classification Of Mammographic Masses Using Circularity Max Metric, A New Method, Tae Keun Heo

Electronic Theses and Dissertations

Breast cancer classification can be divided into two categories. The first category is a benign tumor, and the other is a malignant tumor. The main purpose of breast cancer classification is to classify abnormalities into benign or malignant classes and thus help physicians with further analysis by minimizing potential errors that can be made by fatigued or inexperienced physicians. This paper proposes a new shape metric based on the area ratio of a circle to classify mammographic images into benign and malignant class. Support Vector Machine is used as a machine learning tool for training and classification purposes. The improved …

Product Authentication Using Hash Chains And Printed Qr Codes, Harshith R. Keni

Electronic Theses and Dissertations

This thesis explores the usage of simple printed tags for authenticating products. Printed tags are a cheap alternative to RFID and other tag based systems and do not require specialized equipment. Due to the simplistic nature of such printed codes, many security issues like tag impersonation, server impersonation, reader impersonation, replay attacks and denial of service present in RFID based solutions need to be handled differently. An algorithm that utilizes hash chains to secure such simple tags while still keeping cost low is discussed. The security characteristics of this scheme as well as other product authentication schemes that use RFID …

Obstacle-Aware Wireless Video Sensor Network Deployment For 3d Indoor Space Monitoring, Zhonghui Wang

Electronic Theses and Dissertations

In recent years wireless video sensors networks (WVSNs) have emerged as a leading technology for monitoring 3D indoor space in campus, industrial and medical areas as well as other types of environments. In contrast to traditional sensors such as heat or light sensors often considered with omnidirectional sensing range, the sensing range of a video sensor is directional and can be deemed as a pyramid-shape in 3D. Moreover, in an indoor environment, there are often obstacles such as lamp stands or furniture, which introduce additional challenges and further render the deployment solutions for traditional sensors and 2D sensing field inapplicable …

The Effect Of Hyperparameters In The Activation Layers Of Deep Neural Networks, Clay Lafayette Mcleod

Electronic Theses and Dissertations

Deep neural networks (DNNs), and artificial neural networks (ANNs) in general, have recently received a great amount of attention from both the media and the machine learning community at large. DNNs have been used to produce world-class results in a variety of domains, including image recognition, speech recognition, sequence modeling, and natural language processing. Many of most exciting recent deep neural network studies have made improvements by hardcoding less about the network and giving the neural network more control over its own parameters, allowing flexibility and control within the network. Although much research has been done to introduce trainable hyperparameters …

The Woz Recognizer: A Tool For Understanding User Perceptions Of Sketch-Based Interfaces, Jared Bott

Electronic Theses and Dissertations

Sketch recognition has the potential to be an important input method for computers in the coming years; however, designing and building an accurate and sophisticated sketch recognition system is a time consuming and daunting task. Since sketch recognition is still at a level where mistakes are common, it is important to understand how users perceive and tolerate recognition errors and other user interface elements with these imperfect systems. A problem in performing this type of research is that we cannot easily control aspects of recognition in order to rigorously study the systems. We performed a study examining user perceptions of …

Gesture Assessment Of Teachers In An Immersive Rehearsal Environment, Roghayeh Barmaki

Electronic Theses and Dissertations

Interactive training environments typically include feedback mechanisms designed to help trainees improve their performance through either guided- or self-reflection. When the training system deals with human-to-human communications, as one would find in a teacher, counselor, enterprise culture or cross-cultural trainer, such feedback needs to focus on all aspects of human communication. This means that, in addition to verbal communication, nonverbal messages must be captured and analyzed for semantic meaning. The goal of this dissertation is to employ machine-learning algorithms that semi-automate and, where supported, automate event tagging in training systems developed to improve human-to-human interaction. The specific context in which …

High-Performance Composable Transactional Data Structures, Deli Zhang

Electronic Theses and Dissertations

Exploiting the parallelism in multiprocessor systems is a major challenge in the post ``power wall'' era. Programming for multicore demands a change in the way we design and use fundamental data structures. Concurrent data structures allow scalable and thread-safe accesses to shared data. They provide operations that appear to take effect atomically when invoked individually. A main obstacle to the practical use of concurrent data structures is their inability to support composable operations, i.e., to execute multiple operations atomically in a transactional manner. The problem stems from the inability of concurrent data structure to ensure atomicity of transactions composed from …