Other Computer Engineering Commons^™

Reproducing And Evaluating Charger Surfing: Robustness Of Smartphone Charging-Line Side-Channels, Colby M. Watts

Master's Theses

Smartphones are frequently connected to external, untrusted charging hardware, creating opportunities for side-channel attacks that do not require malware or direct access to device data. Charger Surfing, a recently proposed charging-line power analysis side-channel attack, reported high accuracy in inferring touchscreen input from voltage measurements collected from a smartphone’s charging cable; however, the reproducibility and robustness of these results under different conditions remain unclear. This thesis presents an independent replication and evaluation of Charger Surfing, including the development of an end-to-end data collection pipeline consisting of a modified charging cable, oscilloscope-based recordings, custom Android app, automated trace processing, and convolutional …

Decision Making At Triage Classification Using Svm With Smote Technique, Mehanas Shahul, Pushpalatha Kp

Northeast Journal of Complex Systems (NEJCS)

The efficient functioning of triage gates in overcrowded emergency departments (EDs) occurs in the context of the complex adaptive system (CAS) framework, where diverse system elements – patients, medical personnel, resources, patients’ inflow patterns, and patients themselves – simultaneously and dynamically influence the decision process. This study addresses the automated incorporation of machine learning triage algorithms as part of the system triage process to support automated classified risk-level recognition based on a limited set of vital signs. Patients are dynamically subsumed under high and low-risk categories enhanced by sensitivity, which enables optimal diagnosis and triage response to the critical clinician …

A Machine Learning-Based Apogee Prediction Methodology For Experimental Student Rockets, Price Hamilton Drawdy

Senior Honors Theses

The ability to predict the maximum altitude of a rocket (apogee) in real-time is incredibly useful for collegiate-level competition rockets. This project creates a machine learning-based real-time apogee prediction methodology. Three model types were tested: linear regression, random forest, and a 3-layer multi-layer perceptron (MLP) neural network. These models were trained on a large dataset of simulated flights. All models performed well on simulated test flights, with the linear regression model showing most promise for use on edge compute. More development and real-world testing are necessary to determine how applicable this method is for real-time operation. Nevertheless, this methodology provides …

Performance Analysis Of Sparse Neural Networks In Brain Abnormality Detection, Megan Danh

Honors Undergraduate Theses

Neuroimages have held the capability of revealing to medical professionals patterns for brain abnormalities since their development. However, more recently, these professionals and researchers are looking to use neural networks to identify these brain abnormalities through neuroimages for early detection that would allow more effective treatment. Neuroimage datasets, specifically functional magnetic resonance imaging (fMRI), are extremely large in size. This would result in their processing and training to be computationally expensive, even with smaller neural networks. Fortunately, recent pruning methods have recently emerged, where network weights and neurons are pruned to reduce computational cost without compromising too much accuracy. By …

Low-Resource Ecoacoustic Audio Classification, Enis Berk Coban

Dissertations, Theses, and Capstone Projects

Ecoacoustic monitoring via machine learning enables scalable analysis but is often constrained by labeled data scarcity, particularly in remote regions like the Arctic. This thesis confronts low-resource ecoacoustic audio classification by developing and evaluating complementary machine learning methodologies. We introduce EDANSA, the first publicly available, expert- labeled Arctic dataset of its kind, curated via novel active learning, alongside a baseline CNN. We systematically evaluate transfer learning, showing general audio embeddings effectively bootstrap classifiers for challenging Arctic sounds, significantly outperforming direct label mapping. Optimizing label utility, we investigate standard data augmentation and introduce novel audio data valuation via Shapley values, revealing …

Towards Leveraging Social Media Data For Fostering Collaborations Among Non-Profits, Monazil Chowdhury

LSU Doctoral Dissertations

Nonprofit organizations serve a crucial role in tackling a wide range of significant social, environmental, and economic issues. But it is often hard to get a clear picture of their work because their information is spread out and it is difficult to see how they are collaborating. To address this issue we developed a web-based tool to collect scattered data—from a variety of sources, such as the IRS, social media, and the Census, into one easy-to-use resource. The tool begins by taking IRS records and geocoding each nonprofit’s physical address With its coordinates. It then retrieves census tract information from …

A Comprehensive Analysis Of Recognition Of Hand Gestures Using Machine Learning, Shivani Shivani, Satinder Bal Gupta

Makara Journal of Technology

Hand gestures are a natural means of conveying information and thus, there is an increasing interest in utilizing gestures for communication with computers. This study focuses on systematically reviewing different machine learning algorithms while assessing their working mechanisms and accuracy. Articles were analyzed for comparing the performance of K-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Machines (SVM), Naive Bayes (NB), Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN). In accordance with input data, intricacy of gestures, processing resources, and real-time demands, the study shows that each technique has distinct advantages and disadvantages. RNN showed the best accuracy of …

Role Of Eye-Tracking Technology And Software Algorithms In Enhancing Adhd Detection And Diagnosis: A Systematic Literature Review, Lauren E. Perkins

Honors College Theses

This systematic literature review explores the role of eye-tracking technology and software algorithms in enhancing the detection and diagnosis of ADHD. ADHD, a neurodevelopmental disorder affecting both children and adults, is traditionally diagnosed through behavioral assessments, which may lack objectivity. Recent studies suggest that eye-tracking, specifically focusing on saccades, fixations, and blink rates, offers the potential for more accurate and objective measures of ADHD. The review examines clinical trials, observational studies, and machine learning research to assess the correlation between ADHD and eye movement patterns. Results indicate that individuals with ADHD exhibit distinct eye movement patterns, which can be quantified …

Multimodal Feature Fusion And Machine Learning For Adhd Detection Using Neuroimaging Data, Isabel Pham

Master's Projects

Attention Deficit Hyperactivity Disorder (ADHD) is a common neurodevelopment disorder that can significantly affect a person’s attention, impulse control, and executive function. Currently, the traditional diagnosis method often relies on clinical assessments and observations. However, these methods can be subjective and lead to inconsistencies in diagnosis between individuals. To address this challenge, neuroimaging and machine learning (ML) are promising tools for providing a more objective diagnosis of ADHD. The goal of this project is to apply a multimodal approach in which structural and functional features of specific regions of the brain are used to develop a more accurate and objective …

Identifying Red Sponges On Arms Plates By Preprocessing Images Using Histogram Equalization, Barry Ng

Master's Projects

Sponges play a vital role in marine ecosystems, being the only organisms capable of converting dissolved organic matter (DOM) into particulate organic matter (POM). They provide nutrients for coral reefs to thrive in oligotrophic waters. Autonomous reef monitoring structures (ARMS) are used to measure the biodiversity of coral reefs by simulating the complex cavities inside reef structures. Organisms settle on them and scientists can retrieve them after a period of time for analysis. Images are taken of ARMS plates after they are retrieved. Human analysis is unsuitable for the analysis of ARMS plates due to the huge number of images. …

Designing Customized Loss Functions For Training Deep Neural Networks, Ali Pourramezan Fard

Electronic Theses and Dissertations

This dissertation explores the critical role of loss functions in enhancing the predictive performance of deep machine learning models. Loss functions are an integral element of all the ongoing advances we witness daily in this domain. I design custom loss functions and their impacts on various machine learning tasks, particularly in computer vision.

In the first stage of my research, I aim to improve the prediction performance of deep learning models by providing them with more precise feedback associated with task requirements. This led me to create the concept of assistive loss functions. My first proposed loss function, inspired by …

Impaired Speech Recognition Of Neurological Disorder Persons Using Machine Learning And Deep Learning Techniques, Vishnika Veni S

Theses and Dissertations

Speech Assistive Tools have emerged in recent years to support individuals with cognitive and neurological disorders in the field of assistive technology. People affected by neurological disorders such as autism, stroke, cerebral palsy, dysarthria, Parkinson’s disease, and brain injury often find it difficult to articulate desired sounds, resulting in impaired speech. As the population of impaired speakers continues to increase every year, there is a strong need to develop intelligent speech recognition systems for affected individuals. The primary objective of this research is to develop an Impaired Speech Recognition (ISR) system for the Tamil language. Word Recognition Accuracy (WRA) is …

Machine Learning Security For Tactical Operations, Dr. Denaria Fields, Shakiya A. Friend, Andrew Hermansen, Dr. Tugba Erpek, Dr. Yalin E. Sagduyu

Military Cyber Affairs

Deep learning finds rich applications in the tactical domain by learning from diverse data sources and performing difficult tasks to support mission-critical applications. However, deep learning models are susceptible to various attacks and exploits. In this paper, we first discuss application areas of deep learning in the tactical domain. Next, we present adversarial machine learning as an emerging attack vector and discuss the impact of adversarial attacks on the deep learning performance. Finally, we discuss potential defense methods that can be applied against these attacks.

In-Training Explainability Frameworks To Make Black-Box Machine Learning Models More Explainable., Asuman Cagla Acun

Electronic Theses and Dissertations

Despite ongoing efforts to make black-box machine learning models more explainable, transparent, and trustworthy, there is growing advocacy for using only inherently interpretable models for high-stake decision-making. Post-hoc explanations have recently been criticized for learning surrogate models that may not accurately reflect the actual mechanisms of the original model and for adding computational burden at prediction time. We propose two novel explainability approaches to address these limitations: pre-hoc explainability and co-hoc explainability. These approaches integrate explanations derived from an inherently interpretable white-box model into the learning stage of the black-box model without compromising accuracy. Unlike post-hoc methods, our approach does …

Benchmarking And Enhancing Generalization In Multilingual Speech Emotion Recognition, Mohamed Osman Ismael

Theses and Dissertations

Speech Emotion Recognition (SER) is pivotal in advancing human-computer interaction by enabling machines to understand and respond to human emotions. Despite significant progress with self-supervised learning models, SER systems often struggle with generalization across diverse languages and unseen data distributions, limiting their real-world applicability. This thesis addresses these challenges by first introducing a large-scale benchmark to evaluate the robustness and adaptability of state-of-the-art SER models in both in-domain and out-of-domain settings. The benchmark includes a diverse set of multilingual datasets, emphasizing cross-lingual and out-of-domain evaluations to assess model generalization. Surprisingly, we find that the Whisper model, originally designed for automatic …

A Two-Stage Machine Learning Approach For Fake News Detection And News Article Categorization, Snegdha Adusumilli

Master's Projects

This project employs machine learning techniques to develop a sequential model for detecting and categorizing fake news, aiming to mitigate its proliferation in today's digital landscape. The model operates in two phases: in the first phase, the classification algorithms like Naïve Bayes, XGBoost and Random Forest are used to distinguish between true and false news stories and in the second phase the capabilities of Naïve Bayes, XGBoost, Random Forest, and the Transformer-based BERT (Bidirectional Encoder Representations from Transformers) model are leveraged to further categorize the news into specific topics.

The methodology encompasses several key steps: data acquisition, preprocessing, feature extraction, …

Novel Approach To Music Analysis Using Apache Spark, Nidhi Zare

Master's Projects

Music is one of the most common source of entertainment. Every user has their own taste of music and prefer to listen music that adheres to their taste and mood. There are various categories, called as music genres in which music can be classified. This research project addresses the challenge in music genre classification by using various deep learning models such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Very Deep Convolutional Networks (VGGNet), ResNet and others. The primary objective of this research is to enhance the accuracy of music genre classification using a distributed computing framework Apache Spark. …

Robustness Of Learning Models To Label Flipping Attacks, Sarvagya Bhargava

Master's Projects

In this paper we compare traditional machine learning and deep learning models

trained on a malware dataset when subjected to adversarial attack based on label- flipping. Specifically, we investigate the robustness of different models when faced

with varying percentages of misleading labels, assessing their ability to maintain their accuracy in the face of such adversarial manipulations of the training data. This research aims to provide insights into which models are more robust, in the sense of being better able to resist intentional disruptions to the training data. We find that traditional machine learning models and boosting techniques are more robust …

Multi-Platform Cyberbullying Detection Using Nlp And Machine Learning, Chinmayi Lokeshwar Hegde

Master's Projects

The issue of cyberbullying is growing due to the online anonymity and due to online platforms having less repercussions. This research proposes for proactive measures to detect and prevent such behavior before it reaches the victim. By using data from various social media platforms and employing machine learning techniques, this research proposes an innovative system aimed at identifying and thwarting cyberbullying incidents preemptively. While existing methods have primarily focused on prediction and detection of cyberbullying incidents, there remains a significant gap in research regarding prevention strategies. This project aims to address this gap by leveraging machine learning, natural language processing …

Wall-E: An Autonomous Ai Rover For Precision Agriculture, Simar Ghumman

Master's Projects

Unmanned Ground Vehicles (UGVs) are emerging as a crucial tool in the world of precision agriculture. By working with UGVs equipped with machine learning, we can find solutions to a range of complex agricultural problems. My project, titled “Wall-E: Artificial Intelligence Robot for Precision Agriculture,” focuses on developing a UGV capable of navigating through agriculture fields autonomously while capturing data. Using machine learning, computer vision, and other sensor technologies, Wall-E is capable of estimating the total yield of crops, self-localization, mapping its environment in real time, and avoiding obstacles along its route. The purpose of this project is to automate …

Development Of A Machine Learning System For Irrigation Decision Support With Disparate Data Streams, Eric Wilkening

Department of Agricultural and Biological Systems Engineering: Dissertations, Theses, and Student Research

In recent years, advancements in irrigation technologies have led to increased efficiency in irrigation applications, encompassing the adoption of practices that utilize data-driven irrigation scheduling and leveraging variable rate irrigation (VRI). These technological improvements have the potential to reduce water withdrawals and diversions from both groundwater and surface water sources. However, it is vital to recognize that improved application efficiency does not necessarily equate to increased water availability for future or downstream use. This is particularly crucial in the context of consumptive water use, which refers to water consumed and not returned to the local or sub-regional watershed, representing a …

Detection Of Myofascial Trigger Points With Ultrasound Imaging And Machine Learning, Benjamin Formby

All Theses

Myofascial Pain Syndrome (MPS) is a common chronic muscle pain disorder that affects a large portion of the global population, seen in 85-93% of patients in specialty pain clinics [10]. MPS is characterized by hard, palpable nodules caused by a stiffened taut band of muscle fibers. These nodules are referred to as Myofascial Trigger Points (MTrPs) and can be classified by two states: active MTrPs (A-MTrPs) and latent MtrPs (L-MTrPs). Treatment for MPS involves massage therapy, acupuncture, and injections or painkillers. Given the subjectivity of patient pain quantification, MPS can often lead to mistreatment or drug misuse. A deterministic way …

User Profiling Through Zero-Permission Sensors And Machine Learning, Ahmed Elhussiny

Theses and Dissertations

With the rise of mobile and pervasive computing, users are often ingesting content on the go. Services are constantly competing for attention in a very crowded field. It is only logical that users would allot their attention to the services that are most likely to adapt to their needs and interests. This matter becomes trivial when users create accounts and explicitly inform the services of their demographics and interests. Unfortunately, due to privacy and security concerns, and due to the fast nature of computing today, users see the registration process as an unnecessary hurdle to bypass, effectively refusing to provide …

Sentiment Analysis Of Text And Emoji Data For Twitter Network, Paramita Dey, Soumya Dey

Al-Bahir

Twitter is a social media platform where users can post, read, and interact with 'tweets'. Third party like corporate organization can take advantage of this huge information by collecting data about their customers' opinions. The use of emoticons on social media and the emotions expressed through them are the subjects of this research paper. The purpose of this paper is to present a model for analyzing emotional responses to real-life Twitter data. The proposed model is based on supervised machine learning algorithms and data on has been collected through crawler “TWEEPY” for empirical analysis. Collected data is pre-processed, pruned and …

Detection Of Crypto-Ransomware Attack Using Deep Learning, Muna Jemal

Master of Science in Computer Science Theses

The number one threat to the digital world is the exponential increase in ransomware attacks. Ransomware is malware that prevents victims from accessing their resources by locking or encrypting the data until a ransom is paid. With individuals and businesses growing dependencies on technology and the Internet, researchers in the cyber security field are looking for different measures to prevent malicious attackers from having a successful campaign. A new ransomware variant is being introduced daily, thus behavior-based analysis of detecting ransomware attacks is more effective than the traditional static analysis. This paper proposes a multi-variant classification to detect ransomware I/O …

Predicting Suicide Risk Among Youths Using Machine Learning Methods, Saswati Bhattacharjee

Master's Theses

Suicide is the second leading cause of death among youths in the USA. Although machine learning approaches have provided great potential for predicting suicide risk using survey data, prediction accuracy may not meet the need for clinical diagnosis due to the intrinsic characteristics of datasets. In this study, I perform a comparative study of six classification algorithms including naïve Bayes (NB), logistic regression (LR), multilayer perceptron (MLP), AdaBoost (Ada), random forest (RF), and bagging using YRBSS dataset and investigate the effectiveness of several data handling techniques to improve the overall performance of suicide risk prediction.

The dataset consists of 76 …

Analyzing The Impact Of Automation On Employment In Different Us Regions: A Data-Driven Approach, Thejaas Balasubramanian

Electronic Theses, Projects, and Dissertations

Automation is transforming the US workforce with the increasing prevalence of technologies like robotics, artificial intelligence, and machine learning. As a result, it is essential to understand how this shift will impact the labor market and prepare for its effects. This culminating experience project aimed to examine the influence of computerization on jobs in the United States and answer the following research questions: Q1. What factors affect how likely different jobs will be automated? Q2. What are the possible effects of automation on the US workforce across states and industries? Q3. What are the meaningful predictors of the likelihood of …

Machine-Learning Approaches For Developing An Autograder For High School-Level Cs-For-All Initiatives, Sirazum Munira Tisha

LSU Doctoral Dissertations

Most existing autograders used for grading programming assignments are based on unit testing, which is tedious to implement for programs with graphical output and does not allow testing for other code aspects, such as programming style or structure. We present a novel autograding approach based on machine learning that can successfully check the quality of coding assignments from a high school-level CS-for-all computational thinking course. For evaluating our autograder, we graded 2,675 samples from five different assignments from the past three years, including open-ended problems from different units of the course curriculum. Our autograder uses features based on lexical analysis …

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. …

Interest-Based Recommendation System Using Gmail Topic Modelling, Pranav Ghaskadbi

Master's Projects

Emails are a fundamental part of modern communication. Much of communicative discourse in modern society occurs over email, resulting in personal collections for each mail user which are rich in latent user’s interests. Conventional recommendation systems require historical data of user activity and interactions to derive user interests. The absence of activity and interaction data poses an interesting challenge for generating relevant recommendations for users. We were motivated to investigate approaches to identify user interests in the absence of historical data to generate personalized content recommendations. There is opportunity to derive user interests from email data, which can be used …