Other Computer Engineering Commons^™

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 …

Toward A Generalizable Perceptual Hashing Framework For Image Manipulation Detection, Priyanka Samanta

Dissertations, Theses, and Capstone Projects

This thesis contributes to research in adversarial image manipulation detection. The primary motivation is the increasing need to verify digital images, especially for legal evidence, journalistic proof, or social media content—where manipulated or fabricated images can mislead, defame, or distort reality. A key application and contribution of this work is the development of eWitness, a blockchain application that generates and registers image provenance at capture time to enable independent verification of authenticity. The secret sauce behind the system is SmartHash, a novel and efficient perceptual hashing algorithm designed for real-world deployment in systems like eWitness. Unlike existing algorithms, SmartHash targets …

Correlations Between Song Popularity And Their Audio Features Using Machine Learning, Rong Chen

Dissertations, Theses, and Capstone Projects

This project is an interactive visual project that explores the relationship between audio features and song popularity on Spotify using machine learning techniques. Through the collection of nearly half a million songs and implementation of seven different machine learning models, including Linear Regression, Random Forest, Decision Trees, and Gradient Boosting, I investigated how audio characteristics correlate with a song's popularity ranking. The project utilized MongoDB for data storage, Spotipy for API integration, and Streamlit with Plotly for visualization. This work provides insights into the practical challenges of large-scale music analysis and the relationship between technical audio characteristics and commercial success, …

Communication Challenges In Underwater Wireless Networks: Mac Protocols And Software Solutions, Dmitrii Dugaev

Dissertations, Theses, and Capstone Projects

Underwater wireless networks (UWNs) represent a diverse and intriguing research domain, encompassing a wide array of scientific and industrial applications. This dissertation delves into the communication challenges at the Medium Access Control (MAC) layer within UWNs, stemming from the distinctive signal propagation conditions and the harshness of the deployment environment. The manuscript provides comprehensive coverage of key aspects of UWNs, including potential applications, communication protocols, methodologies employed in such networks, and existing software solutions that facilitate simulation, emulation, and real testbed scenarios for underwater research endeavors. Furthermore, this research introduces innovative software and communication solutions designed to facilitate the seamless …

Multi-Perspective Analysis For Derivative Financial Product Prediction With Stacked Recurrent Neural Networks, Natural Language Processing And Large Language Model, Ethan Lo

Dissertations, Theses, and Capstone Projects

This study developed a multi-perspective, AI-powered model for predicting E-Mini S&P 500 Index Futures prices, tackling the challenging market dynamics of these derivative financial instruments. Leveraging FinBERT for analysis of Wall Street Journal data alongside technical indicators, trader positioning, and economic factors, my stacked recurrent neural network built with LSTMs and GRUs achieves significantly improved accuracy compared to single sub-models. Furthermore, ChatGPT generation of human-readable analysis reports demonstrates the feasibility of using large language models in financial analysis. This research pioneers the use of stacked RNNs and LLMs for multi-perspective financial analysis, offering a novel blueprint for automated prediction and …

Stand-Up Comedy Visualized, Berna Yenidogan

Dissertations, Theses, and Capstone Projects

Stand-up comedy has become an increasingly popular form of comedy in the recent years and comedians reach audiences beyond the halls they are performing through streaming services, podcasts and social media. While comedic performances are typically judged by how 'funny' they are, which could be proxied by the frequency and intensity of laughs through the performance, comedians also explore untapped social issues and provoke conversation, especially in this age where interaction with artists goes beyond their act. It is easy to see commonalities in the topics addressed in comedians’ work such as relationships, race and politics.This project provides an interactive …

Deception Detection Across Domains, Languages And Modalities, Subhadarshi Panda

Dissertations, Theses, and Capstone Projects

With the increase of deception and misinformation especially in social media, it has become crucial to develop machine learning methods to automatically identify deception. In this dissertation, we identify key challenges underlying text-based deception detection in a cross-domain setting, where we do not have training data in the target domain. We analyze the differences between domains and as a result develop methods to improve cross-domain deception detection. We additionally develop approaches that take advantage of cross-lingual properties to support deception detection across languages. This involves the usage of either multilingual NLP models or translation models. Finally, to better understand multi-modal …

Happiness And Policy Implications: A Sociological View, Sarah M. Kahl

Dissertations, Theses, and Capstone Projects

The World Happiness Report is released every year, ranking each country by who is “happier” and explaining the variables and data they have used. This project attempts to build from that base and create a machine learning algorithm that can predict if a country will be in a “happy” or “could be happier” category. Findings show that taking a broader scope of variables can better help predict happiness. Policy implications are discussed in using both big data and considering social indicators to make better and lasting policies.

A Machine Learning Approach To Predicting The Onset Of Type Ii Diabetes In A Sample Of Pima Indian Women, Meriem Benarbia

Dissertations, Theses, and Capstone Projects

Type II diabetes is a disease that affects how the body regulates and uses sugar (glucose) as a fuel. This chronic disease results in too much sugar circulating in the bloodstream. High blood sugar levels can lead to circulatory, nervous, and immune systems disorders. Machine learning (ML) techniques have proven their strength in diabetes diagnosis. In this paper, we aimed to contribute to the literature on the use of ML methods by examining the value of a number of supervised machine learning algorithms such as logistic regression, decision tree classifiers, random forest classifiers, and support vector classifiers to identify factors …

Efficient Protocols For Multi-Party Computation, Tahereh Jafarikhah

Dissertations, Theses, and Capstone Projects

Secure Multi-Party Computation (MPC) allows a group of parties to compute a join function on their inputs without revealing any information beyond the result of the computation. We demonstrate secure function evaluation protocols for branching programs, where the communication complexity is linear in the size of the inputs, and polynomial in the security parameter. Our result is based on the circular security of the Paillier's encryption scheme. Our work followed the breakthrough results by Boyle et al. [9; 11]. They presented a Homomorphic Secret Sharing scheme which allows the non-interactive computation of Branching Programs over shares of the secret inputs. …

A Purely Defeasible Argumentation Framework, Zimi Li

Dissertations, Theses, and Capstone Projects

Argumentation theory is concerned with the way that intelligent agents discuss whether some statement holds. It is a claim-based theory that is widely used in many areas, such as law, linguistics and computer science. In the past few years, formal argumentation frameworks have been heavily studied and applications have been proposed in fields such as natural language processing, the semantic web and multi-agent systems. Studying argumentation provides results which help in developing tools and applications in these areas. Argumentation is interesting as a logic-based approach to deal with inconsistent information. Arguments are constructed using a process like logical inference, with …

In Search Of Homo Sociologicus, Yunqi Xue

Dissertations, Theses, and Capstone Projects

The subject of this dissertation is to build an epistemic logic system that is able to show the spreading of knowledge and beliefs in a social network that contains multiple subgroups. Epistemic logic is the study of logical systems that express mathematical properties of knowledge and belief. In recent years, there have been increasing number of new epistemic logic systems that are focused on community properties such as knowledge and belief adoption among friends.

We are interested in revisable and actionable social knowledge/belief that leads to a large group action. Instead of centralized coordination, bottom-up approach is our focus. We …

Scale Up Bayesian Network Learning, Xiannian Fan

Dissertations, Theses, and Capstone Projects

Bayesian networks are widely used graphical models which represent uncertain relations between the random variables in a domain compactly and intuitively. The first step of applying Bayesian networks to real-word problems is typically building the network structure. Optimal structure learning via score-and-search has become an active research topic in recent years. In this context, a scoring function is used to measure the goodness of fit of a structure to given data, and the goal is to find the structure which optimizes the scoring function. The problem has been viewed as a shortest path problem, and has been shown to be …