Computer Engineering Commons^™

Analysis Of Attention Mechanisms In Box-Embedding Systems, Jeffrey Sardina Jeffrey Sardina, Callie Sardina, John Kelleher, Declan O’Sullivan

Conference papers

Large-scale Knowledge Graphs (KGs) have recently gained considerable research attention for their ability to model the inter- and intra- relationships of data. However, the huge scale of KGs has necessitated the use of querying methods to facilitate human use. Question Answering (QA) systems have shown much promise in breaking down this human-machine barrier. A recent QA model that achieved state-of-the-art performance, Query2box, modelled queries on a KG using box embeddings with an attention mechanism backend to compute the intersections of boxes for query resolution. In this paper, we introduce a new model, Query2Geom, which replaces the Query2box attention mechanism with …

Action Classification In Human Robot Interaction Cells In Manufacturing, Shakra S.M. Mehak, Maria Chiara Leva, John Kelleher, Michael Guilfoyle

Conference papers

Action recognition has become a prerequisite approach to fluent Human-Robot Interaction (HRI) due to a high degree of movement flexibility. With the improvements in machine learning algorithms, robots are gradually transitioning into more human-populated areas. However, HRI systems demand the need for robots to possess enough cognition. The action recognition algorithms require massive training datasets, structural information of objects in the environment, and less expensive models in terms of computational complexity. In addition, many such algorithms are trained on datasets derived from daily activities. The algorithms trained on non-industrial datasets may have an unfavorable impact on implementing models and validating …

Comparative Adjudication Of Noisy And Subjective Data Annotation Disagreements For Deep Learning, Scott David Williams

Browse all Theses and Dissertations

Obtaining accurate inferences from deep neural networks is difficult when models are trained on instances with conflicting labels. Algorithmic recognition of online hate speech illustrates this. No human annotator is perfectly reliable, so multiple annotators evaluate and label online posts in a corpus. Labeling scheme limitations, differences in annotators' beliefs, and limits to annotators' honesty and carefulness cause some labels to disagree. Consequently, decisive and accurate inferences become less likely. Some practical applications such as social research can tolerate some indecisiveness. However, an online platform using an indecisive classifier for automated content moderation could create more problems than it solves. …

Enhancing Graph Convolutional Network With Label Propagation And Residual For Malware Detection, Aravinda Sai Gundubogula

Browse all Theses and Dissertations

Malware detection is a critical task in ensuring the security of computer systems. Due to a surge in malware and the malware program sophistication, machine learning methods have been developed to perform such a task with great success. To further learn structural semantics, Graph Neural Networks abbreviated as GNNs have emerged as a recent practice for malware detection by modeling the relationships between various components of a program as a graph, which deliver promising detection performance improvement. However, this line of research attends to individual programs while overlooking program interactions; also, these GNNs tend to perform feature aggregation from neighbors …

Effective Systems For Insider Threat Detection, Muhanned Qasim Jabbar Alslaiman

Browse all Theses and Dissertations

Insider threats to information security have become a burden for organizations. Understanding insider activities leads to an effective improvement in identifying insider attacks and limits their threats. This dissertation presents three systems to detect insider threats effectively. The aim is to reduce the false negative rate (FNR), provide better dataset use, and reduce dimensionality and zero padding effects. The systems developed utilize deep learning techniques and are evaluated using the CERT 4.2 dataset. The dataset is analyzed and reformed so that each row represents a variable length sample of user activities. Two data representations are implemented to model extracted features …

Fuzzing Php Interpreters By Automatically Generating Samples, Jacob S. Baumgarte

Browse all Theses and Dissertations

Modern web development has grown increasingly reliant on scripting languages such as PHP. The complexities of an interpreted language means it is very difficult to account for every use case as unusual interactions can cause unintended side effects. Automatically generating test input to detect bugs or fuzzing, has proven to be an effective technique for JavaScript engines. By extending this concept to PHP, existing vulnerabilities that have since gone undetected can be brought to light. While PHP fuzzers exist, they are limited to testing a small quantity of test seeds per second. In this thesis, we propose a solution for …

Human Tracking Function For Robotic Dog, Andrew Sharkey

Williams Honors College, Honors Research Projects

With the increase the increase in automation and humans and robots working side by side, there is a need for a more organic way of controlling robots. The goal of this project is to create a control system for Boston dynamics robotic dog Spot that implements human tracking image software to follow humans using computer vision as well as using hand tracking image software to allow for control input through hand gestures.

Improving Developers' Understanding Of Regex Denial Of Service Tools Through Anti-Patterns And Fix Strategies, Sk Adnan Hassan, Zainab Aamir, Dongyoon Lee, James C. Davis, Francisco Servant

Department of Electrical and Computer Engineering Faculty Publications

Regular expressions are used for diverse purposes, including input validation and firewalls. Unfortunately, they can also lead to a security vulnerability called ReDoS (Regular Expression Denial of Service), caused by a super-linear worst-case execution time during regex matching. Due to the severity and prevalence of ReDoS, past work proposed automatic tools to detect and fix regexes. Although these tools were evaluated in automatic experiments, their usability has not yet been studied; usability has not been a focus of prior work. Our insight is that the usability of existing tools to detect and fix regexes will improve if we complement them …

An Empirical Study Of Pre-Trained Model Reuse In The Hugging Face Deep Learning Model Registry, Wenxin Jiang, Nicholas Synovic, Matt Hyatt, Taylor R. Schorlemmer, Rohan Sethi, Yung-Hsiang Lu, George K. Thiruvathukal, James C. Davis

Department of Electrical and Computer Engineering Faculty Publications

Deep Neural Networks (DNNs) are being adopted as components in software systems. Creating and specializing DNNs from scratch has grown increasingly difficult as state-of-the-art architectures grow more complex. Following the path of traditional software engineering, machine learning engineers have begun to reuse large-scale pre-trained models (PTMs) and fine-tune these models for downstream tasks. Prior works have studied reuse practices for traditional software packages to guide software engineers towards better package maintenance and dependency management. We lack a similar foundation of knowledge to guide behaviors in pre-trained model ecosystems.

In this work, we present the first empirical investigation of PTM reuse. …

The Role Of Data Analytics To Address Water Stress In Africa, Khalil Beladda

ICT

Water stress, a global concern transcending geographical boundaries, significantly impacts the African continent. Affecting one in three people in Africa, sustainable water management is imperative for ecological and human welfare. This research emphasizes the pivotal role of data analytics in addressing water stress challenges in Africa and beyond.

Applying Machine Learning To Biological Status (Qvalues) From Physio-Chemical Conditions Of Irish Rivers, Raúl Martín Sánchez

ICT

This thesis evaluates and optimises a variety of predictive models for assessing biological classification status, with an emphasis on water quality monitoring. Grounded in previous pertinent studies, it builds on the findings of (Arrighi and Castelli, 2023) concerning Tuscany’s river catchments, highlighting a solid correlation between river ecological status and parameters like summer climate and land use. They achieved an 80% prediction precision using the Random Forest algorithm, particularly adept at identifying "good" ecological conditions, leveraging a dataset devoid of chemical data.

Recurrent Neural Networks For Flash Gdp Estimates In Ireland: A Comparison With Traditional Econometric Methods, Justin Flannery

ICT

GDP is the single most important barometer for the health of an economy. It’s an important input into the decision making processes of government, industry and state institutions such as central banks. To be useful as an indicator, GDP estimates need to be both timely and accurate. To meet the needs of users, many national statistical institutes publish early or flash estimates of GDP which are produced within 30 days after the end of a quarter. Given the long lags involved in the data collection processes which feed into GDP estimates, these flash estimates are often largely model based. Within …

Unlocking The Pragmatics Of Emoji: Evaluation Of The Integration Of Pragmatic Markers For Sarcasm Detection, Niamh Farnham

ICT

Emojis have become an integral element of online communications, serving as a powerful, under-utilised resource for enhancing pragmatic understanding in NLP. Previous works have highlighted their potential for improvement of more complex tasks such as the identification of figurative literary devices including sarcasm due to their role in conveying tone within text. However present state-of-the-art does not include the consideration of emoji or adequately address sarcastic markers such as sentiment incongruence. This work aims to integrate these concepts to generate more robust solutions for sarcasm detection leveraging enhanced pragmatic features from both emoji and text tokens. This was achieved by …

Smart Power Outlet, Kyrolos Melek, Joseph M. Garro, Ethan G. Frese, Madison Britton

Williams Honors College, Honors Research Projects

With an ever-increasing demand on the electrical grid and an increase in electrical consumer goods per household, effective power management and monitoring for home users is a must. The objective of this project is to create a consumer outlet and companion phone application. The power outlet will measure power draw on each socket and report these measurements via an embedded system, wirelessly, to the application. The application will then allow for setting custom power draws at which an individual socket will switch off if exceeded and will display how much a given outlet or individual socket is costing per unit …

Leveraging Ansible To Locate Network Devices, Elijah Hadden

Williams Honors College, Honors Research Projects

This report demonstrates the collection of Layer 2 and Layer 3 network device information using Ansible and explains the limitations identified performing such a procedure. Maintaining dynamic documentation of network information using automation would benefit network security with improvements to attack surface visibility and cyberattack incident response time. After the procedure was conducted, it was determined that Ansible alone is incapable of producing the intended information base due to the way it handles numerical variables. This report identifies limitations in Ansible’s ability to manipulate large quantities of network information and necessitates further discussion regarding what Ansible can and should be …

Space Force Design Project, Emily Greene, Ashton Orosa, Julia Patek, Nathan Doty

Williams Honors College, Honors Research Projects

The objective of our research project is to develop a lab testbed composed of a curved surface to represent a spacecraft hull, a mobile robot equipped with repair tools, and a robotic arm equipped with a laser 3D scanner. This project is part of a larger grant to the University of Akron from Space Force and Air Research Labs. The lab testbed developed in this project will be used to assist in creating and testing a software and algorithm to inspect and repair spacecraft while in orbit. The project will involve researching spacecraft hulls to create an accurate simulation bed, …

Detection Of Various Dental Conditions On Dental Panoramic Radiography Using Faster R-Cnn, Shih Lun Chen, Tsung Yi Chen, Yi Cheng Mao, Szu Yin Lin, Ya Yun Huang, Chiung An Chen, Yuan Jin Lin, Mian Heng Chuang, Patricia Angela R. Abu

Department of Information Systems & Computer Science Faculty Publications

The dental panoramic radiograph (DPR) is a pivotal diagnostic tool in dentistry. However, despite the growing prevalence of artificial intelligence (AI) across various medical domains, manual methods remain the prevailing means of interpreting DPR images. This study aims to introduce an advanced identification system for detecting seven dental conditions in DPR images by utilizing Faster R-CNN. The primary objectives are to enhance dentists' efficiency and evaluate the performance of various CNN models as foundational training networks. This study contributes significantly to the field in several notable ways. Firstly, including a Butterworth filter in the training process yielded an approximately 7% …

Performance Evaluation Of Face Mask Detection For Real-Time Implementation On An Rpi, Ivan George L. Tarun, Vidal Wyatt M. Lopez, Pamela Anne C. Serrano, Patricia Angela R. Abu, Rosula Reyes, Ma. Regina Justina Estuar

Department of Information Systems & Computer Science Faculty Publications

Mask-wearing remains to be one of the primary protective measures against COVID-19. To address the difficulty of manual compliance monitoring, face mask detection models considerate of both frontal and angled faces were developed. This study aimed to test the performance of the said models in classifying multi-face images and upon running on a Raspberry Pi device. The accuracies and inference speeds were measured and compared when inferencing images with one, two, and three faces and on the desktop and the Raspberry Pi. With an increasing number of faces in an image, the models’ accuracies were observed to decline, while their …

Towards The Development Of A Blockchain System For Philippine Government Processes For Enhanced Transparency And Verifiability, Karlo Angelo F. Cabugwang, Raphael Christen K. Enriquez, Bienvenido E. Villabroza, Christian E. Pulmano

Department of Information Systems & Computer Science Faculty Publications

There are cases of corruption and fraud within the Philippine government that have gone under the radar, often due to a lack of transparency and verifiability. The objective of this study is to prototype a blockchain network that can run a government process as a decentralized application such that it can enhance transparency and verifiability in the public sector. This can be accomplished by identifying a government process that would be converted into a decentralized application. One of these processes would be converted into a decentralized application. Afterwards, a blockchain framework should be identified one which can create a public …

Unsupervised-Based Distributed Machine Learning For Efficient Data Clustering And Prediction, Vishnu Vardhan Baligodugula

Browse all Theses and Dissertations

Machine learning techniques utilize training data samples to help understand, predict, classify, and make valuable decisions for different applications such as medicine, email filtering, speech recognition, agriculture, and computer vision, where it is challenging or unfeasible to produce traditional algorithms to accomplish the needed tasks. Unsupervised ML-based approaches have emerged for building groups of data samples known as data clusters for driving necessary decisions about these data samples and helping solve challenges in critical applications. Data clustering is used in multiple fields, including health, finance, social networks, education, and science. Sequential processing of clustering algorithms, like the K-Means, Minibatch K-Means, …

Anomaly Detection In Multi-Seasonal Time Series Data, Ashton Taylor Williams

Browse all Theses and Dissertations

Most of today’s time series data contain anomalies and multiple seasonalities, and accurate anomaly detection in these data is critical to almost any type of business. However, most mainstream forecasting models used for anomaly detection can only incorporate one or no seasonal component into their forecasts and cannot capture every known seasonal pattern in time series data. In this thesis, we propose a new multi-seasonal forecasting model for anomaly detection in time series data that extends the popular Seasonal Autoregressive Integrated Moving Average (SARIMA) model. Our model, named multi-SARIMA, utilizes a time series dataset’s multiple pre-determined seasonal trends to increase …

Path-Safe: Enabling Dynamic Mandatory Access Controls Using Security Tokens, James P. Maclennan

Browse all Theses and Dissertations

Deploying Mandatory Access Controls (MAC) is a popular way to provide host protection against malware. Unfortunately, current implementations lack the flexibility to adapt to emergent malware threats and are known for being difficult to configure. A core tenet of MAC security systems is that the policies they are deployed with are immutable from the host while they are active. This work looks at deploying a MAC system that leverages using encrypted security tokens to allow for redeploying policy configurations in real-time without the need to stop a running process. This is instrumental in developing an adaptive framework for security systems …

Data-Driven Strategies For Disease Management In Patients Admitted For Heart Failure, Ankita Agarwal

Browse all Theses and Dissertations

Heart failure is a syndrome which effects a patient’s quality of life adversely. It can be caused by different underlying conditions or abnormalities and involves both cardiovascular and non-cardiovascular comorbidities. Heart failure cannot be cured but a patient’s quality of life can be improved by effective treatment through medicines and surgery, and lifestyle management. As effective treatment of heart failure incurs cost for the patients and resource allocation for the hospitals, predicting length of stay of these patients during each hospitalization becomes important. Heart failure can be classified into two types: left sided heart failure and right sided heart failure. …

Understanding And Enhancing The Efficiency And Efficacy Of Machine Learning-Assisted Software Vulnerability Detection, Daniel J. Grahn

Browse all Theses and Dissertations

As our world has become dependent upon software for nearly every aspect of modern society, software security has followed as an essential feature. The first line of defense against vulnerabilities is secure coding. While today’s programmers are carefully taught secure coding best practices, they can make mistakes or intentionally introduce vulnerable code. The traditional backstop to human errors and insider threats is the adoption of automated security analysis tools. These analysis tools have limitations. Static analysis suffers from high false positive rates that may cause annoyance and complacency among developers. Dynamic analysis can be difficult to set up and very …

Machine Learning For Biosensors, Gayathri Anapanani

Graduate Theses, Dissertations, and Problem Reports (ETD)

Biosensors have become increasingly popular as diagnostic tools due to their ability to detect and quantify biological analytes in a wide range of applications. With the growing demand for faster and more reliable biosensing devices, machine learning has become a valuable tool in enhancing biosensor performance. In this report, we review recent progress in the application of machine learning to biosensors. We discuss the potential benefits of using machine learning in biosensors, including improved sensitivity, selectivity, and accuracy. We also discuss the various machine learning techniques that have been applied to biosensors, including data preprocessing, feature extraction, and classification and …

Implementing Test Automation With Selenium Webdriver, Ramana Inturi

Graduate Theses, Dissertations, and Problem Reports (ETD)

Many software programs, such as applications for designing, modeling, simulating, and analyzing systems, are now commonly available as web-based applications. The testing of such sophisticated web applications is highly challenging and can be extremely tedious and error-prone if done manually. Recently automation tools have become increasingly used for testing web-based applications, as they minimize human involvement and repetitive work.

For this problem report project, we have built and implemented an automation testing framework for web applications. The project specifically uses a tool called Selenium WebDriver, which has been used to develop the testing framework. By using this framework, testers may …

A Flexible Photonic Reduction Network Architecture For Spatial Gemm Accelerators For Deep Learning, Bobby Bose

Theses and Dissertations--Electrical and Computer Engineering

As deep neural network (DNN) models increase significantly in complexity and size, it has become important to increase the computing capability of specialized hardware architectures typically used for DNN processing. The major linear operations of DNNs, which comprise the fully connected and convolution layers, are commonly converted into general matrix-matrix multiplication (GEMM) operations for acceleration. Specialized GEMM accelerators are typically employed to implement these GEMM operations, where a GEMM operation is decomposed into multiple vector-dot-product operations that run in parallel. A common challenge that arises in modern DNNs is the mismatch between the matrices used for GEMM operations and the …

Application Of Conventional Feedforward And Deep Neural Networks To Power Distribution System State Estimation And State Forecasting, James Paul Carmichael

Theses and Dissertations--Electrical and Computer Engineering

Classical neural networks such as feedforward multilayer perceptron models (MLPs) are well established as universal approximators and as such, show promise in applications such as static state estimation in power transmission systems. This research investigates the application of conventional neural networks (MLPs) and deep learning based models such as convolutional neural networks (CNNs) and long short-term memory networks (LSTMs) to mitigate challenges in power distribution system state estimation and forecasting based upon conventional analytic methods. The ability of MLPs to perform regression to perform power system state estimation will be investigated. MLPs are considered based upon their promise to learn …

Learning Outcomes And Learner Satisfaction: The Mediating Roles Of Self-Regulated Learning And Dialogues, Sean Eom, Nicholas Jeremy Ashill

Journal of International Technology and Information Management

The interdependent learning process is regarded as a crucial part of e-learning success, but it has been largely ignored in e-learning empirical research. Grounded in constructivist and social constructivist theory, we present and test an e-learning success model consisting of eight e-learning critical success factors (CSF) derived from constructivist and social constructivist models. Three hundred seventy-two on-line students from a Midwestern university in the United States participated in the survey. The data collected from the survey was used to examine the partial least squares structural equation model. The results highlight the importance of self-regulated learning and dialogical processes to explain …

Acceptance Of Interoperable Electronic Health Record (Ehrs) Systems: A Tanzanian E-Health Perspective, Emmanuel Mbwambo, Herman Mandari

Journal of International Technology and Information Management

The study assessed factors that influence the acceptance of interoperable electronic Health Records (EHRs) Systems in Tanzania Public Hospitals. The study applied a hybrid model that combined the Technology Acceptance Model (TAM) and Technology-Organization-Environment (TOE). Snowball sampling technique was applied and a total of 340 questionnaires were distributed to selected clinics, polyclinics and hospitals, of which 261 (77%) received questionnaires were considered to be valid and reliable for subsequent data analysis. IBM SPSS software version 27.0 was employed for data analysis. Findings indicated that relative advantage, compatibility, management support, organizational competency, training and education, perceived ease of use, perceived usefulness, …