Computer Engineering Commons^™

Understanding The Limits Of Deep Packet Inspection For Network Traffic Classification, Herman Ramey

Open Access Theses & Dissertations

We present our human network application labeling system that contributes a new level of distinction between the network traffic that should be labeled from the network traffic that should not be labeled. This distinction improves the label accuracy of the training data set produced from the human labeled data and will subsequently improve the performance of supervised machine learning classifiers used for network traffic classification. This system also allows for the human network user to label traffic, with little effort, in a manner consistent with normal network usage, i.e., no need for a contrived experiment. Lastly, we use human supplied …

An Edge Computing System With Amd Xilinx Fpga Ai Customer Platform For Advanced Driver Assistance System, Tsun Kuang Chi, Tsung Yi Chen, Yu Chen Lin, Ting Lan Lin, Jun Ting Zhang, Cheng Lin Lu, Shih Lun Chen, Kuo Chen Li, Patricia Angela R. Abu

Department of Information Systems & Computer Science Faculty Publications

The convergence of edge computing systems with Field-Programmable Gate Array (FPGA) technology has shown considerable promise in enhancing real-time applications across various domains. This paper presents an innovative edge computing system design specifically tailored for pavement defect detection within the Advanced Driver-Assistance Systems (ADASs) domain. The system seamlessly integrates the AMD Xilinx AI platform into a customized circuit configuration, capitalizing on its capabilities. Utilizing cameras as input sensors to capture road scenes, the system employs a Deep Learning Processing Unit (DPU) to execute the YOLOv3 model, enabling the identification of three distinct types of pavement defects with high accuracy and …

Ai-Powered Information Retrieval In Meeting Records And Transcripts Enhancing Efficiency And User Experience, Srushti Nitin Ghadge

Theses and Dissertations

This study compares the traditional search methods, which is to search from video recordings of the meetings by moving the slider back and forth or by keyword search in transcripts versus integrated AI video plus transcript search. Based on the previous test results, we introduced some human-centric design features to the AI and built a new enhanced AI search tool for information retrieval. For search technique efficiency testing, the method had two set of experiments. The first results of the experiment showed that AI-based search algorithms were more accurate and faster than conventional search approaches. Participants were also happier with …

Particle Swarm Optimization For Training Quadrotor Pid Controller, Eric Xavier Rodriguez

Theses and Dissertations

The objective of this research is to establish a fundamental approach to tuning PID (Proportional-Integral-Derivative) parameters for a simulated quadrotor drone. Implementing a PID controller for autonomous flight provides a straightforward and efficient method for monitoring and correcting robotic movement based on the robot's current state. However, applying a PID approach to a quadrotor's flight controller poses challenges, such as assigning multiple parameters to control an inherently under-actuated system. This includes the need to find optimal parameter values that reduce the likelihood of large overshoots and lengthy adjustment times. Ineffectively tuning PID parameters can have detrimental effects on autonomously …

Developing Resilient Defense Strategies Against Pheromone-Based Attacks In Foraging Robot Swarms, Ryan A. Luna

Theses and Dissertations

This thesis delves into the security of stochastic pheromone-based foraging algorithms within swarm robotic systems, a subset of foraging algorithms distinguished by their reliance on probabilistic decision-making mechanisms inspired by the natural world. Such algorithms face vulnerabilities in stigmergic communication that threaten to disrupt swarm operations. This research investigates these vulnerabilities, presenting two distinct contributions.

The first contribution examines the implementation of quarantine strategies as a defensive measure to isolate and mitigate the impact of fake resource attacks. By simulating these attacks, this study quantitatively assesses their detrimental effects on swarm efficiency and explores the efficacy …

Understanding Timing Error Characteristics From Overclocked Systolic Multiply-Accumulate Arrays In Fpgas, Andrew S. Chamberlin

All Graduate Theses and Dissertations, Fall 2023 to Present

Artificial Intelligence (AI) is one of the biggest fields of research for computer hardware right now. Hardware accelerators are chips (such as graphics cards) that are purpose built to be the best at a specific type of operation. AI hardware accelerators are a growing field of research. Part of hardware in general is a digital clock that controls the pace at which computations occur. If this clock runs too quickly, the hardware won't have enough time to finish its computation. We call that a timing error. This paper focuses on studying the characteristics of timing errors in a small custom …

Automatic Speech Recognition For Air Traffic Control Using Convolutional Lstm, Sakshi Nakashe

Electronic Theses, Projects, and Dissertations

The need for automatic speech recognition in air traffic control is critical as it enhances the interaction between the computer and human. Speech recognition helps to automatically transcribe the communication between the pilots and the air traffic controllers, which reduces the time taken for administrative tasks. This project aims to provide improvement to the Automatic Speech Recognition (ASR) system for air traffic control by investigating the impact of convolution LSTM model on ASR as suggested by previous studies. The research questions are: (Q1) Comparing the performance of ConvLSTM with other conventional models, how does ConvLSTM perform with respect to recognizing …

Analyzing An In-Line Compression Management System For Improved Performance In A High-Performance Computing Environment, Steven Platt

All Theses

High-performance computing (HPC) has enabled advancements in computation speed and resource cost by utilizing all available server resources and using parallelization for speedup. This computation scheme encourages simulation model development, massive data collection, and AI computation models, all of which store and compute on massive amounts of data. Data compression has enhanced the performance of storing and transferring this HPC application data to enable acceleration, but the benefits of data compression can also be transferred to the active allocated memory used by the application. In-line compression is a compression method that keeps the application memory compressed in allocated memory, decompressing …

Deep Reinforcement Learning Of Variable Impedance Control For Object-Picking Tasks, Akshit Lunia

All Theses

The increasing deployment of robots in industries with varying tasks has accelerated the development of various control frameworks, enabling robots to replace humans in repetitive, exhaustive, and hazardous jobs. One critical aspect is the robots' interaction with their environment, particularly in unknown object-picking tasks, which involve intricate object weight estimations and calculations when lifting objects. In this study, a unique control framework is proposed to modulate the force exerted by a manipulator for lifting an unknown object, eliminating the need for feedback from a force/torque sensor. The framework utilizes a variable impedance controller to generate the required force, and an …

Defining And Labeling Traversable Space In A Forested Environment, James Nguyen

All Theses

This thesis investigates the problem of identifying traversable terrain in outdoor conditions. We are motivated by research in recent years toward identifying drivable space for the purpose of developing autonomous vehicles. Our motivating application is similar but also different. We envision a “Hiker Helper” that assists humans with dismounted navigation in forested terrain. A common challenge in this type of environment is identifying a viable path for moving through terrain that is congested with trees, bushes, other flora, and natural obstacles that would make navigation difficult. We envision training an artificial intelligence (AI) model to automatically analyze images of this …

Hand Movement Analysis For Surgical Suturing Skill Assessment, Amir Mehdi Shayan

All Theses

To enhance patient safety, surgical education is increasingly incorporating simulation for formative skills assessment and training. However, many standardized assessment tools rely on human raters for performance assessment, which is resource-intensive and subjective. Simulators that provide automated and objective metrics from sensor data can address this limitation. This thesis presents an instrumented bench suturing simulator, patterned after the Clock Face (CF) radial suturing model from the Fundamentals of Vascular Surgery (FVS), for automated and objective assessment of open suturing skills by particularly focusing on biomechanical analysis of hand movements. For this research, 97 participants (35 attending surgeons and fellows, 32 …

Multi-Domain Secure Dds Networks For Aerial And Ground Vehicle Communications, Daniel Pendleton

All Theses

none

Recommender System Design And Multi-Channel Pricing: Personalization Strategies For Online Platforms, Hao Zhang

Dissertations and Theses Collection (Open Access)

The advancement of mobile technology and rising consumer demands have contributed to the unprecedented growth of online platforms. In online platforms, recommender systems connect with multistakeholders who have different interests. Designing recommender systems to balance the benefit of multistakeholders is important for these platforms. In addition, price is an important factor influencing consumers’ purchase decisions. An increasing number of online platforms introduce multiple sales channels. Optimizing multiple-channel prices is vital for these platforms. Thus, this thesis designs multistakeholder recommender systems and multi-channel pricing strategies for online platforms through the following two works.

The first work focuses on designing multistakeholder recommender …

Experiment Development And Validation Of A Granular Jamming Robotic Gripper, Jacob R. Dowd

UNLV Theses, Dissertations, Professional Papers, and Capstones

A granular jamming gripper (GJG) is widely known as a Universal Gripper because of the wide range of objects that it can grasp and the simplicity of control, design, and manufacturing. Despite multitude of research improving the GJG, here, we focus on the base version of the GJG and attempt to glean the range of objects that it may reliably grasp. Despite the limited range of objects, which were a sphere, rectangular prism, and cylinder, we gleaned geometric properties as it relates to successful and unsuccessful grasping. This was based on the two types of testing: push and pull testing …

Attention Guided Data Augmentation For Improving The Classification Performance Of Vision Transformers., Nada Baili

Electronic Theses and Dissertations

For over a decade, Deep Neural Networks (DNNs) have been rapidly progressing and achieving great success, forming a robust foundation of state of the art machine learning algorithms that impacted various domains. The advances in data acquisition and processing have undeniably played a major role in these breakthroughs. Data is a crucial component in building successful DNNs, as it enables machine learning models to optimize complex architectures, necessary to perform certain difficult tasks. However, acquiring large-scale data sets is not enough to learn robust models with generalizable features. Instead, an ideal training set should be diverse enough and contain enough …

Multithreaded Applications On The Heterogeneous Research Computing Environment., Sungbo Jung

Electronic Theses and Dissertations

Bioinformatics is a domain that has experienced rapid research growth in recent years, as evidenced by the increasing number of articles in biomedical databases such as PubMed, which adds over a million publications every year. However, this also poses a challenge for researchers who need to find relevant citations for their work. Therefore, developing efficient indexing and searching methods for text data is crucial for Bioinformatics. One key technique for information retrieval is document inversion, which involves creating an inverted index to enable efficient searching through vast collections of text or documents. This Ph.D. research aims to design the research …

Simulating And Training Autonomous Rover Navigation In Unity Engine Using Local Sensor Data, Christopher Pace

Senior Honors Theses

Autonomous navigation is essential to remotely operating mobile vehicles on Mars, as communication takes up to 20 minutes to travel between the Earth and Mars. Several autonomous navigation methods have been implemented in Mars rovers and other mobile robots, such as odometry or simultaneous localization and mapping (SLAM) until the past few years when deep reinforcement learning (DRL) emerged as a viable alternative. In this thesis, a simulation model for end-to-end DRL Mars rover autonomous navigation training was created using Unity Engine, using local inputs such as GNSS, LiDAR, and gyro. This model was then trained in navigation in a …

4-Channel Spatially Multiplexed Communication System In Single-Core Optical Fibers, Ce Su

Theses and Dissertations

This dissertation delves into exploring and advancing spatial domain / space division multiplexing (SDM) technologies within single-core optical fibers, a frontier in optical fiber communications poised to meet the burgeoning global demand for data transmission. At the heart of this research is the pursuit to significantly enhance the capacity and efficiency of optical fiber communication systems without necessitating additional fiber infrastructure. This work unveils a new paradigm in optical fiber communications characterized by a pioneering 4-channel SDM system through a meticulous process encompassing theoretical modeling, computational simulations, design innovations, and rigorous experimental validations. Theoretical contributions include the development of refined …

Design And Application Of Smart Systems To Address Analytical Problems, Lucas B. Ayres

All Dissertations

This dissertation is a multidisciplinary effort that integrates low-cost analytical instrumentation, redox chemistry, and artificial intelligence to overcome existing limitations in the fields of wearable sensing technology, Deep Eutectic Solvents (DES), and antioxidant chemistry. The overall goal behind each implemented strategy is to enhance the accuracy, efficiency, and accessibility of analytical processes and technologies. A general overview of the thesis, along with the research outcomes is included in Chapter One. The theoretical framework of this dissertation is presented in Chapter Two. Chapter Three describes the development of a wearable platform (sensor and instrumentation) to rapidly detect (~20 minutes) S. aureus …

Non-Vacuous Generalization Bounds For Adversarial Risk In Stochastic Neural Networks, Mustafa Waleed, Liznerski Philipp, Antoine Ledent, Wagner Dennis, Wang Puyu, Kloft Marius

Research Collection School Of Computing and Information Systems

Adversarial examples are manipulated samples used to deceive machine learning models, posing a serious threat in safety-critical applications. Existing safety certificates for machine learning models are limited to individual input examples, failing to capture generalization to unseen data. To address this limitation, we propose novel generalization bounds based on the PAC-Bayesian and randomized smoothing frameworks, providing certificates that predict the model’s performance and robustness on unseen test samples based solely on the training data. We present an effective procedure to train and compute the first non-vacuous generalization bounds for neural networks in adversarial settings. Experimental results on the widely recognized …

Geometric Multi-Resolution Analysis Across Signals, Images, And Networks, Felicia Schenkelberg

Dartmouth College Master’s Theses

This research delves into the transformative potential of Geometric Multi-Resolution Analysis (GMRA) as a robust tool for dimensionality reduction and data analysis in the context of high-dimensional graphs. Statistical techniques for classification have historically been tailored for scenarios wherein the number of observations significantly exceeds the number of features, a paradigm characteristic of low-dimensional datasets. However, recent advancements in technologies have ushered in a transformative era in data collection practices across diverse domains, resulting in the acquisition of extensive feature measurements. As a result of this shift, datasets have transitioned into a high-dimensional realm wherein the number of features significantly …

Cloud Computing Integration Into Mixed-Reality: Physical To Abstraction, Yassine Chahid, Patrick Slattery

Publications and Research

This research evaluates the progression of cloud computing and mixed-reality technologies, and to identify how these technologies influence advancements in the latter. Both cloud computing and mixed reality have significantly impacted the IT field and the services available to the public and various institutions. Cloud computing provides a valuable way to process information or allocate computational resources on otherwise limited hardware. Augmented or virtual reality hardware would greatly benefit from this by offloading resource-intensive tasks to other machines. The research methodology involves analyzing essential components of both innovations, divided into multiple categories. These components range from physical, hardware-based elements to …

Detection Of Deficiencies And Data Analysis Of Bridge Members With Deep Convolutional Neural Networks, Bennett Jackson

Department of Civil and Environmental Engineering: Dissertations, Theses, and Student Research

Concrete cracks and structural steel corrosion are two of the most common defects in bridges. Quantifying and classifying these defects provide bridge inspectors and engineers with valuable data for assessing deterioration levels. However, the bridge inspection process is typically a subjective, time intensive, and tedious task, as defects can be overlooked or in locations not easily accessible. Previous studies have investigated deep learning-based inspection methods, implementing popular models such as Mask R-CNN and U-Net. The architectures of these models offer certain advantages depending on the required task. This thesis aims to evaluate and compare Mask R-CNN and U-Net regarding their …

Sliding Markov Decision Processes For Dynamic Task Planning On Uncrewed Aerial Vehicles, Trent Wiens

Department of Mechanical and Materials Engineering: Dissertations, Theses, and Student Research

Mission and flight planning problems for uncrewed aircraft systems (UASs) are typically large and complex in space and computational requirements. With enough time and computing resources, some of these problems may be solvable offline and then executed during flight. In dynamic or uncertain environments, however, the mission may require online adaptation and replanning. In this work, we will discuss methods of creating MDPs for online applications, and a method of using a sliding resolution and receding horizon approach to build and solve Markov Decision Processes (MDPs) in practical planing applications for UASs. In this strategy, called a Sliding Markov Decision …

Investigating Factors Influencing Blockchain Adoption In Saudi Healthcare Data Management, Noura Mohammad Alkhalifah

Theses and Dissertations

Blockchain technology can potentially address security and privacy issues concerning the collection, storage, and sharing of healthcare data. However, its adoption within the healthcare sector is nascent in Saudi Arabia. This underutilization prompted our investigation into the determinants influencing blockchain adoption, intending to fully empower the Saudi healthcare sector to leverage blockchain capabilities. To achieve this, an extensive literature review was conducted to identify the pivotal factors encompassing technology, organization, and environment (TOE) that affect the successful implementation of blockchain technologies in managing healthcare data within the Saudi context. Utilizing the TOE framework, this study formulated three hypotheses concerning the …

Deep Learning Using Vision And Lidar For Global Robot Localization, Brett E. Gowling

Master's Theses

As the field of mobile robotics rapidly expands, precise understanding of a robot’s position and orientation becomes critical for autonomous navigation and efficient task performance. In this thesis, we present a snapshot-based global localization machine learning model for a mobile robot, the e-puck, in a simulated environment. Our model uses multimodal data to predict both position and orientation using the robot’s on-board cameras and LiDAR sensor. In an effort to minimize localization error, we explore different sensor configurations by varying the number of cameras and LiDAR layers used. Additionally, we investigate the performance benefits of different multimodal fusion strategies while …

A Smart Hybrid Enhanced Recommendation And Personalization Algorithm Using Machine Learning, Aswin Kumar Nalluri

Electronic Theses, Projects, and Dissertations

In today’s age of streaming services, the effectiveness and precision of recommendation systems are crucial in improving user satisfaction. This project introduces the Smart Hybrid Enhanced Recommendation and Personalization Algorithm (SHERPA) a cutting-edge machine learning approach aimed at transforming how movie suggestions are made. By combining Term Frequency Inverse Document Frequency (TF-IDF) for content based filtering and Alternating Squares (ALS) with Weighted Regularization for filtering SHERPA offers a sophisticated method for delivering tailored recommendations.

The algorithm underwent evaluation using a dataset that included over 50 million ratings from 480,000 Netflix users encompassing 17,000 movie titles. The performance of SHERPA was …

Cultural Awareness Application, Bharat Gupta

Electronic Theses, Projects, and Dissertations

In an increasingly interconnected global landscape, cultural awareness and competency have become indispensable skills for individuals and organizations alike. This paper introduces a pioneering cultural awareness application, grounded in the Cultural Orientation Model—a comprehensive framework devised by Dr. Walker [8]to guide individuals in understanding, appreciating, and effectively engaging with diverse cultures. The application encompasses ten primary dimensions, each representing fundamental aspects of social life shared by members of any socio-cultural environment. Through a combination of cultural education, interactive learning, guidance on cultural etiquette, and integration of cultural events, the application aims to foster empathy, tolerance, and effective cross-cultural communication skills. …

Automated Brain Tumor Classifier With Deep Learning, Venkata Sai Krishna Chaitanya Kandula

Electronic Theses, Projects, and Dissertations

Brain Tumors are abnormal growth of cells within the brain that can be categorized as benign (non-cancerous) or malignant (cancerous). Accurate and timely classification of brain tumors is crucial for effective treatment planning and patient care. Medical imaging techniques like Magnetic Resonance Imaging (MRI) provide detailed visualizations of brain structures, aiding in diagnosis and tumor classification^[8].

In this project, we propose a brain tumor classifier applying deep learning methodologies to automatically classify brain tumor images without any manual intervention. The classifier uses deep learning architectures to extract and classify brain MRI images. Specifically, a Convolutional Neural Network (CNN) …

Effectiveness Of Cnn-Lstm Models Used For Apple Stock Forecasting, Ethan White

Electronic Theses, Projects, and Dissertations

This culminating experience project investigates the effectiveness of convolutional neural networks mixed with long short-term memory (CNN-LSTM) models, and an ensemble method, extreme gradient boosting (XGBoost), in predicting closing stock prices. This quantitative analysis utilizes recent AAPL stock data from the NASDAQ index. The chosen research questions (RQs) are: RQ1. What are the optimal hyperparameters for CNN-LSTM models in stock price forecasting? RQ2. What is the best architecture for CNN-LSTM models in this context? RQ3. How can ensemble techniques like XGBoost effectively enhance the predictions of CNN-LSTM models for stock price forecasting?

The research questions were answered through a thorough …