Digital Commons Network^™

Research On Economic Dispatching Strategy Of Chp Units Based On Srl, Xin Wang, Chenggang Cui, Xiangxiang Wang, Ping Zhu

Journal of System Simulation

Abstract: In addressing the challenge of the DRL algorithm in the optimization of combined heat and power (CHP) units, lacking safety and stability guarantees, a scheduling optimization method based on SRL is proposed. Utilizing Dymola platform, a district heating system model is constructed with the CHP unit as the heat source. A MDP model for the economic dispatching of CHP units is designed, incorporating control barrier functions (CBF) to guide safe exploration in DRL. Simulation results show that the CBF-DRL method, in complex and nonlinear district heating systems, not only accelerates the convergence of DRL algorithms but also efficiently utilizes …

Mobile Robot Path Planning Based On Search-Step Optimized A* Algorithm, Die Yu, Baizhong Bao, Yan Si, Jian Duan, Xiaobin Zhan, Tielin Shi

Journal of System Simulation

Abstract: A search-step optimized A* algorithm is proposed to address the issues with the traditional A* algorithm in robot path planning tasks, such as the high time consumption in large-scale high-resolution maps and the poor paths qualitys. Based on the cubic Hermite curve, a set of search steps (the path edges connecting the current node to its successors) is constructed, which can match the size of the robot and satisfy the dynamic constraints of the robot. More accurate cost functions are established based on the length and maximum absolute curvature value of the curve. Experimental results show that compared with …

Signal Timing Optimization Via Reinforcement Learning With Traffic Flow Prediction, Ming Xu, Jinye Li, Dongyu Zuo, Jing Zhang

Journal of System Simulation

Abstract: In response to the existing reinforcement learning-based traffic signal control methods that do not consider the changing trends in traffic flow, leading to congestion and inability to adapt to complex and variable road conditions, we propose a traffic signal timing optimization reinforcement learning method based on flow prediction. A phase timing amplitude control model is introduced. This model analyzes the spatiotemporal characteristics of historical traffic data to predict the flow for the next time slot and calculates a reasonable range for phase timing based on the prediction results. The H-PPO algorithm is employed to control the signal phase while …

An Intelligent Tracking Control Method For Unmanned Vehicles With Time-Varying Disturbances, Jie Huang, Jie Huang

Journal of System Simulation

Abstract: An intelligent policy iteration tracking control method is proposed for the tracking control problem with bounded time-varying disturbances. An adaptive disturbance compensator is designed to counteract the bounded disturbance and guarantee the validity of the Hamilton-Jacobi-Bellman (HJB) equation. An identifier network is proposed to estimate the unknown vehicle dynamics, and a new HJB equation is derived using the reconstructed identifier tracking error. An online optimal tracking control strategy for unmanned vehicles is obtained in the state of identifier estimation with the assistance of actor-critic network. Based on Lyapunov theory, it is demonstrated that the identifier tracking error, identifier approximation …

Capability Dependency Analysis Based On Kill Chain And Fdna, Yushuai Wang, Guangya Si

Journal of System Simulation

Abstract: To better support the operation SoS analysis, deeply analyze the impact of dependency relationship during mission accomplishment, and accurately grasp the deep logic of SoS capability generation, the capability dependency analysis method based on the kill chain and function dependency network analysis(FDNA) is proposed. Combined with the analysis of the characteristics of the capability dependency relationship, the kill chain closure and the kill web formation process are abstracted from the perspective of operational interaction, a capability dependency network modeling method for the SoS is proposed, and a specific process covering the identification of capability dependency, calculation of operability, solving …

Seg-Swin: A Dual-Attention Transformer Model For Advanced Amd Classification And Lesion Detection Using Color Fundus Imaging, Niveen Nasr El-Den, Mohamed Elsharkawy, Ibrahim Saleh, Ali H. Mahmoud, Mohammed Ghazal, Ashraf Khalil, Ashraf Sewelam, Hani Mahdi, Ayman El-Baz

All Works

Age-related macular degeneration (AMD) is a prevalent retinal disorder in the elderly, often leading to significant vision impairment. The diagnosis of AMD is confirmed through various medical imaging modalities, with color fundus photography (CFP) being a primary tool. The detection and staging of AMD-severity depend on several factors, including the number and size of drusen, the presence of pigmentary changes, geographic atrophy, and neovascularization, all of which are identifiable through CFP. In this study, we introduce an innovative dual-vision transformer-based network designed to automatically detect AMD and classify its severity into either dry AMD or wet AMD using CFP. Early …

Breaking Down Terminology Of Clojure Error Messages For Beginner Programmers, John Walbran, Jaydon Stanislowski, Tristan Kalvoda

Undergraduate Research Symposium 2025

The Clojure programming language has educational potential for beginner programmers due to its clean, simple syntax and its strong focus on functional programming, an important aspect of CSci education. However, one weakness of Clojure lies in its error messages, which are messages that programmers receive when a program goes wrong. The terminology and shorthands used to convey necessary information for understanding the error are often confusing to novices. The issue is exacerbated by the fact that the error messages are phrased in terms of the underlying programming language – Java – which beginner programmers may typically be unfamiliar with. A …

Elder Fraud Metrics And Preventative Measures Of Chesapeake, Virginia, Joey J. Whitmore Jr.

Cybersecurity Undergraduate Research Showcase

Geriatric crime continues to escalate in the digital era, where older individuals are disproportionately being targeted because of their low digital literacy and high susceptibility to online frauds. In this paper, we examine the breadth of elder fraud in Chesapeake, Virginia using FBI Internet Crime Complaint Center (IC3) data and state-level cybersecurity initiatives and survey responses. Older adults aged 60 and up have reported losses of over $3.4 billion in 2023 alone, underscoring the importance of proactive measures. It assesses the public awareness from traditional and AI-based perspectives revealing significant gaps in digital safety literacy and fraud reporting mechanism among …

The Hidden Carbon Footprint Of Ai Models: Gpu-Aware Carbon Modeling, Youzhi Li

Undergraduate Research Symposium 2025

The rapid growth of AI technology has sparked transformative innovations but also increased carbon emissions. Recent research found that computer systems' carbon emissions are shifting from operational carbon to embodied carbon, but they did not fully capture the rapidly evolving AI landscape. Most recent research focused on operational carbon, neglecting the long-term environmental impact of embodied carbon. We found two gaps that persist in recent research. First, current carbon modeling focused on Central Processing Units (CPUs), neglecting the carbon modeling of Graphical Processing Units (GPUs). Second, it focused on primary components, neglecting significant contributions from peripheral components to the embodied …

Analyzing Musical Emotions: A Multi-Dataset Approach To Sentiment And Mood Classification In Songs, Mahad Syed

SPARK Symposium Presentations

Music evokes a wide range of emotions, yet most music recommendation systems focus on sound and listening patterns rather than the meaning of lyrics. This project enhances lyric-based emotion recognition by applying Natural Language Processing (NLP) and Machine Learning (ML) to classify song lyrics into emotional categories.

I used eight datasets from Kaggle, including collections of lyrics, emotion labels, and audio features, providing a strong foundation for analysis. Our approach combines traditional NLP techniques (like TF-IDF and Word2Vec) with advanced deep learning models (such as BERT and XLNet) to classify lyrics into categories like happy, sad, angry, calm, romantic, and …

Knowledge Distillation For Efficient Object Detection: Toward Scalable And Deployable Vision Models, Qizhen Lan

All ETDs from UAB

Object detection is a critical component of autonomous driving, requiring real-time, robust perception to ensure safety. However, state-of-the-art deep neural network object detectors typically incur high computational cost and memory footprint, hindering their deployment in resource-constrained environments such as self-driving vehicles. This dissertation addresses the need for efficient yet accurate detectors by leveraging knowledge distillation (KD), a model compression technique that transfers knowledge from a high-capacity teacher model to a lightweight student model. While KD has seen success in image classification, its application to object detection poses unique challenges due to multiple instances per image and complex output structures. To …

Grp-082 Ready Cluster One: Optimizing Film Success With Data Science, Mohsin Md Abdul Karim, Joseph Richardson

C-Day Computing Showcase

This study presents a novel approach to predicting and optimizing screenplay investments by combining graph theory and finite mixture modeling (FMM) techniques. We construct a k-partite graph representing movies, genres, subgenres, production companies, directors, actors, and directors of photography, to explore the interconnectedness between these entities. Using FMM, we identify clusters within budget tiers, enabling a deeper understanding of how similar films perform based on their creative team and production characteristics. By balancing profit potential with risk-adjusted profit, the model suggests the most viable budget tiers for unproduced screenplays. This approach incorporates confidence intervals and evaluates the accuracy of budget …

Gc-074 Real-Time Object Detection, Rohit Malik

C-Day Computing Showcase

This project explores the implementation of real-time object detection using the You Only Look Once (YOLO) architecture. Leveraging its speed and accuracy, we developed a system capable of identifying and localizing multiple objects within live video streams. Our implementation focused on optimizing YOLO's performance for real-time applications, specifically addressing the trade-off between speed and accuracy. We employed a pre-trained YOLO model and fine-tuned it on a custom dataset tailored to specific object classes. This fine-tuning process aimed to enhance the model's ability to recognize objects in our target environment. The system was implemented using Python and the OpenCV library, enabling …

Gc-079 Nibbleai, Ryan Deem, Zeynep Birgili, Austin Cook

C-Day Computing Showcase

Ever looked into your fridge or pantry and wondered, “What can I make with this?” NibbleAI is a mobile app designed to solve exactly that. Using artificial intelligence, the app identifies ingredients from user-uploaded images and suggests recipes based on what’s available. Built with React Native and powered by a DenseNet169 model for image recognition, NibbleAI seamlessly analyzes photos and returns curated recipe ideas — all within a few taps. This intuitive approach helps users reduce food waste, save time, and get creative with the ingredients they already have.

Grm-076 Assessing The Performance Of Intelligent Agents In Visual Food Recognition Relative To Manual Data Entry, El Arbi Belfarsi, Henry Ekeocha

C-Day Computing Showcase

Accurate dietary assessment remains a critical yet time-consuming task in health and nutrition monitoring. This study benchmarks the macronutrient estimation capabilities of three intelligent vision agents: GPT Vision, Claude, and Gemini against manually logged food data. We unify two distinct datasets: MenuMatch, annotated by a professional nutritionist, and CGMacros, populated through user entries on MyFitnessPal. After flattening and cleaning both datasets, we first assess each model’s performance in calorie estimation. GPT Vision outperforms the others with the lowest percentage error 13.83% and is subsequently used to benchmark the macro estimations of Claude and Gemini. While Claude shows higher carbohydrate and …

Grm-109 Quantum Machine Learning For Science And Engineering Research, Andrew Polisetty

C-Day Computing Showcase

This research project aims to understand and explore the practical applications of Quantum Machine Learning (QML) in solving real-world challenges. By comparing classical machine learning models such as Support Vector Machines (SVM), Neural Networks, Logistic Regression, and Naive Bayes, with their quantum counterparts. Quantum Support Vector Machines (QSVM), Quantum Neural Networks (QNN), Quantum Logistic Regression (QLR), Quantum Deep Neural Networks (QDNN), and Hybrid Quantum Models, we gain hands-on experience in advanced machine learning techniques. The project cover diverse domains including cybersecurity, healthcare, industrial engineering, energy management, and supply chain optimization. Each part of project involves working with real-world datasets, preprocessing, …

Grp-021 Shap-Explainable Image-To-Topology Regression, Charles Fanning

C-Day Computing Showcase

We evaluated whether deep regression models predicting vectorized topological features (in the form of persistence landscapes) actually learn the underlying persistent homology of the image. A DenseNet-121 is trained to regress 300-dimensional persistence landscapes from grayscale scene images. Using SHAP, we evaluate the contribution of pixels in the original images to the persistence landscapes. Across all six classes, SHAP-feature overlap is consistently lower than the baseline, implying that DenseNet may not be truly learning the underlying persistent homology.

Uc-040 Security Lookup Interface Project, Alhasan Mohsen, Alejandro Albarran, Ethan Lan, Jerrat Jester

C-Day Computing Showcase

The "Security Lookup Interface" capstone project aims to create a streamlined tool for COX's cybersecurity team, enabling analysts to efficiently perform IP address and hostname lookups while providing actionable, data-driven insights to enhance security investigations. The project will develop a user-friendly interface that simplifies the lookup process, allowing cybersecurity analysts to quickly retrieve relevant data and make informed decisions during security investigations. One of the key features of the tool is its seamless integration with both internal APIs and external resources. This integration will ensure that analysts have quick and easy access to valuable information, minimizing manual effort and enabling …

Uc-049 From Forecast To Fortune: Portfolio Optimization And Prediction, Nia Taylor

C-Day Computing Showcase

This project explores the intersection of time series forecasting and portfolio optimization to support data-driven investment strategies. Historical price data from 30 individual stocks was analyzed using two forecasting models: ARIMA and Prophet. Each model’s performance was evaluated using key accuracy metrics, including Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE), and Mean Directional Accuracy (MDA). Results showed that ARIMA performed better on error-based metrics, while Prophet excelled at predicting directional trends. In parallel, historical return data was used to construct optimized portfolios using Modern Portfolio Theory. Two strategies were implemented: one minimizing overall volatility and another maximizing …

Ur-001 Large Language Model Enabled Mental Health App Recommendations Using Structured Datasets, Kris Prasad

C-Day Computing Showcase

The increasing use of large language models (LLMs) in mental health support neces-sitates detailed evaluation of their recommendation capabilities. This study compares four modern LLMs—GPT-4o, Claude 3.5 Sonnet, dataset-enhanced Gemma 2, and dataset-enhanced GPT-3.5-Turbo—in recommending mental health applications. We constructed a structured dataset of 55 mental health apps using RoBERTa-based sentiment analysis and keyword similarity scoring, focusing on depression, anxiety, ADHD, and insomnia. Standard LLMs emonstrated inconsistent accuracy and often relied on outdated or generic information. In contrast, our retrieval-augmented generation (RAG) pipeline enabled lower-cost models to achieve 100% accuracy, compared to baseline models (GPT-4o at 45% and Claude at …

Uc-116 Robot Tactics, John Anderson

C-Day Computing Showcase

Robot Tactics is a first person strategic shooter made in Unity where the player takes control of an agent who fights off bodyguards who are chasing him while using Robots to detour them

Ur-094 Aistudy: Using Ai To Study Ai, Mason Pederson

C-Day Computing Showcase

Interactive AI studying tool or AIStudy is a flask-based web-app which enables users to quickly search, save, and study scientific papers. AIStudy streamlines the literature review process by utilizing large language models (LLMs) allowing for users to engage with research in a creative and interactive way. To begin with a user searches up papers using the arXiv API and PyMuPDF for scraping the contents. These are saved to a user database managed by SQL Alchemy. The user can then ask a chatbot about one or more papers at a time through Ollama’s API in order to produce Retrieval-Augmented Generated (RAG) …

Gc-008 Medivault: An Ai-Powered Secure Medical Image Sharing Platform, Henry Ekeocha, Reginald Calixte, Dhruv Patel

C-Day Computing Showcase

MediVault is a secure, cloud-native platform that empowers patients and healthcare providers to upload, view, and share medical images like X-rays and MRIs with confidence. Built using Next.js and Node.js, and deployed on AWS Free Tier (S3, RDS, KMS), the system implements role-based access, two-factor authentication, and end-to-end encryption to ensure privacy and HIPAA compliance. MediVault features an intuitive interface and integrates AI-enhanced automation to streamline metadata tagging and detect potential anomalies in scans. Designed for scalability, usability, and compliance, this project showcases real-world expertise in full-stack cloud development and healthcare cybersecurity.

Gc-025 Secure Medical Image Sharing Platform – Medshare, Toyese Adenuga, Richard Lateu

C-Day Computing Showcase

The Secure Medical Image Sharing Platform is a cloud-based solution that ensures secure upload, management, and sharing of medical images, adhering to HIPAA and GDPR. It utilizes advanced encryption protocols, role-based access control (RBAC), and audit trails to safeguard patient data. The platform's user-friendly interface facilitates seamless interaction between patients and healthcare providers, enabling the use of AI tools for diagnostic support.

Gc-058 Personalized Wellness Recommendations, Varshini Yaganti

C-Day Computing Showcase

A health recommendation system using machine learning, built on Hadoop, Spark, and HDFS, represents a significant advancement in personalized healthcare. This project aims to leverage big data technologies to process and analyze vast amounts of medical data across a distributed computing environment, utilizing at least three virtual machines. The background of this project lies in the increasing prevalence of chronic diseases and the growing volume of health-related data collected by healthcare providers. The motivation for this project stems from several key factors. Firstly, traditional healthcare systems often struggle to provide personalized recommendations due to the sheer volume and complexity of …

Gc-123 Deep Learning-Based Skin Cancer Detection, Andrew Polisetty, Akshay Krishna Varma Buddharaju, Siri Yellu

C-Day Computing Showcase

Skin cancer is increasingly becoming a severe health problem globally today, but early detection is essential to enhance survival rates. Nonetheless, conventional diagnosis relies largely on visual examinations by dermatologists, which can be subjective and time-consuming. This research examines the application of deep learning for the automation of skin cancer detection based on dermoscopic images from the HAM10000 dataset. The models VGG19, DenseNet121 and ResNet152 will be trained and evaluated, with class mbalance addressed using data augmentation strategies. The outputs will demonstrate the applicability of deep learning to improve skin cancer diagnosis. Classification optimization using ensemble modeling and its improved …

Grm-011 Non-Invasive Convolution-Based Coronary Artery Blood Pressure Prediction, Rene Lisasi

C-Day Computing Showcase

The primary objective of this proposal is to develop an innovative technique for determining the functional significance of coronary artery lesions in patients with coronary artery disease (CAD) and evaluate its utility for clinical decision-making using coronary computed tomography angiography (CCTA).

Grm-022 Exploring Coronavirus 2019 Datasets With Convolutional Neural Networks, Ryan Deem, Sandeep Kumar Vupputuri, Tyler Hood, Droan Patel, Stephen Jacobs

C-Day Computing Showcase

Over the past several decades, the healthcare sector has increased its data creation velocity at an astonishing rate. More doctors and patients have access to real-time imaging technology, which leads to earlier detection and diagnosis for a variety of diseases. In this project, we will explore several datasets that were gathered by various health organizations during the Coronavirus 2019 (COVID-19) pandemic. We will leverage big data analytics techniques and neural network modeling to gain deeper insights into the differentiated diagnosis of COVID-19.

Grm-041 Ai-Driven Analysis Of Openalg Curriculum: Mapping Ai Competencies Across Georgia’S Higher Education Landscape, Michelle Ilechukwu, Iveren Leghemo, Timothy Ngeny

C-Day Computing Showcase

This project investigates the presence of artificial intelligence (AI) competencies across Georgia’s higher education curriculum using university course catalogs as the primary data source, supplemented by OpenALG materials. We applied large language models, including OpenAI’s ChatGPT and embedding APIs, to analyze over 34,000 courses summarizing content, classifying AI relevance, and mapping to global frameworks (AI4K12 and UNESCO). Techniques such as topic clustering, semantic similarity analysis, and geographic distribution mapping were used to uncover patterns in AI integration. Findings reveal that AI content is concentrated in computing disciplines and research universities, with limited coverage in community colleges, MSIs, and non-technical fields. …

Grm-043 Performance Assessment Of Deepseek Versus Bard And Chatgpt In Detecting Alzheimer’S Dementia, Muhammad Awais Arshad

C-Day Computing Showcase

Alzheimer’s disease is a growing public health issue due to its progressive nature and increasing prevalence. Large language models (LLMs) offer promising avenues for non-invasive cognitive assessment through natural language understanding. In this study, we evaluate DeepSeek’s general-purpose model V3 and reasoning-enhanced R1 variant—for identifying Alzheimer’s dementia (AD) and Cognitively Normal (CN) individuals using transcripts derived from spontaneous speech. Two baseline prompting strategies (zero-shot, chain-of-thought ) were applied to both model types and an additional query (self-consistency prompting) was applied to assess better predictions. Accuracy was the primary performance metric. When positively identifying AD, the general-purpose DeepSeek V3 model produced …