Physical Sciences and Mathematics Commons^™

Ur-49 Coronary Artery Segmentation Using Convolutional Neural Network, Connor Bell, Brenda Nyiam, Daron L Pracharn

C-Day Computing Showcase

The project contributes to the advancement of medical imaging technology by overcoming the challenges associated with segmenting coronary arteries from ICA images. By leveraging deep learning algorithms, the system can effectively extract coronary arteries with high accuracy, providing valuable information for CAD diagnosis and treatment planning. Accurate and efficient coronary artery segmentation can improve the workflow of cardiologists and enhance the quality of patient care. A robust automated segmentation model could potentially reduce the time and resources required for manual annotation by experienced cardiologists, leading to cost savings and increased efficiency in clinical settings. Additionally, the developed model could be …

Ur-70 Faster Inequivalence Testing Using Robustness, Emily G Jackson

C-Day Computing Showcase

We propose a new method for quickly testing the inequivalence of two Boolean functions, when one function is represented as an ordered binary decision diagram (OBDD), and the other is represented in conjunctive normal form (CNF). Our approach is based on a notion of classifier robustness from the fields of explainable AI (XAI) and adversarial machine learning. In particular, we show that two Boolean functions that are very similar in terms of their truth values, can be very different in terms of their robustness, which in turn, provides a witness to their inequivalence. A more efficient approach to inequivalence testing …

Ur-78 Transforming Game Play: A Comparative Study Of Cnn And Transformer Based Q-Networks In Reinforcement Learning, William A Stigall

C-Day Computing Showcase

In this study we investigate the performance of Deep Q-Networks utilizing Convolutional Neural Networks (CNNs) and Transformer architectures across 3 different Atari Games. The advent of DQNs have significantly advanced Reinforcement Learning, enabling agents to directly learn optimal policy from high dimensional sensory inputs from pixel or RAM data. While CNN based DQNs have been extensively studied and deployed in various domains Transformer based DQNs are relatively unexplored. Our research aims to fill this gap by benchmarking the performance of both DCQNs and DTQNs across the Atari games' Asteroids, Space Invaders and Centipede. Our research finds that our Transformer Agent …

Ur-94 Emohydra: Multimodal Emotion Classification Using Heterogenous Modality Fusion, William A Stigall

C-Day Computing Showcase

Affective computing is a field of growing importance, as human society becomes more integrated with machines. Human feelings are both complex and multi-modal, expressed through various methods and nuances in behavior. In this work we introduce EmoHydra, a multi-modal model created through the fusion of three top-level models fine-tuned on text, vision, and speech respectively. Despite heterogenous heads performing well on the unseen data, as well as generalizing well to other benchmarks, logit concatenation proves to be ineffective at predicting Multimodal data, therefore we implement Multi-Head Attention as our fusion mechanism.

Gc-67 Automatic Daily Financing Headlines Collection, Storage, Analysis And Presentation, Sahithi Atimamula, Jasminn O Evans, Julian Reyes, Victoria Castro, Krista G Settle

C-Day Computing Showcase

The "Automatic Daily Financing Headlines Collection, Storage, Analysis, and Presentation" project aims to streamline the process of gathering financial headlines from financial news sources, storing them systematically, performing analysis, and presenting the insights in a user-friendly format. This automation project is designed to provide timely and relevant financial information to users interested in staying informed about market trends, economic news, and financial events. By automating the entire workflow, users can stay informed, make data-driven decisions, and navigate the dynamic landscape of financial news with ease. The project embodies the fusion of automation, data analytics, and user-centric design to create a …

Gmr-90 Digimindready: Enhancing Military Readiness With Edge Ai-Driven Wellness, Education, And Digital Discipline Through Mhealth Innovation., Md Mehedi Hasan, Nafisa Anjum

C-Day Computing Showcase

Military personnel often need to operate in high-stakes situations. Combating such volatile missions primarily includes control over cognitive overload, reckless mindset, and maintaining concentration amid distractions to sustain operational effectiveness. Military training significantly focuses on human performance, which benefits military readiness. However, the 21st century has introduced unanticipated challenges, such as adverse effects of excessive screen time, external distractions, and over-reliance on technology to the US military, on top of existing issues like anxiety and emotional stability, adversely impacting military readiness and decreasing quality of life. A strategic investigation into these issues and the advancement of effective tools to address …

Gmr-47 A Two-Stage Prediction Model For House Prices, Nguyen Thi Binh Nguyen, Brandon Bell, Syanthan Reddy Ravula, Hari Krishna Thota

C-Day Computing Showcase

Predicting house prices is a challenging task that researchers from various fields (economics, statistics, politics, etc.) have attempted to answer. An accurate house prediction is useful not only to policymakers to improve their policies, but also to help sellers and buyers in the real estate market make well- informed decisions. Commonly, prediction models are trained on the whole dataset. However, as Azimlu et al [1] suggested, such models might not perform very well on dispersed data. They propose a new approach which first divides the whole dataset into smaller clusters, and then each cluster would be trained with an appropriate …

Gmr-72 Ai-Based Discourse Analysis System (Adas) For Improved Stem Education, Varun Gottam

C-Day Computing Showcase

In the rapidly evolving fields of Artificial Intelligence and Natural Language Processing, significant opportunities have emerged to transform educational practices. Discourse analysis, particularly in science education, plays a critical role in fostering scientific thinking among students. However, the manual application of tools like the Classroom Discourse Analysis Tool is resource-intensive and impractical on a large scale. This abstract proposes the development of an AI-based Discourse Analysis System tailored for educational settings, designed to automate and enrich the analysis of classroom discourse. Leveraging the latest in Artificial Intelligence and Natural Language Processing, this web-based application will provide teachers nationwide with the …

Ur-87 Adversarial Patch Attack In Deep Learning Based Remote Sensing Object Detection Model, Kyle Bratcher

C-Day Computing Showcase

Advancements in the field of machine learning have led to object detection systems that can approach or even improve upon human performance. Based on deep learning, these systems play a crucial role in many aspects, and continue to be improved on and see expanded adoption. However, these systems are vulnerable to adversarial attacks that rely on targeted noise to spoof detection. Researchers have applied this concept to increase real world adversarial performance by restricting this noise to a patch that can be placed on new images to disrupt object detection. Previous research has focused on patches applied to person recognition …

Evaluating Introductory Computer Science Labs In The Presence Of Ai Tools, Nicholas Snow, Devin Chaimberlain, Abigail Pitcairn, Benjamin Sweeney

Thinking Matters Symposium

This study explores the resistance of introductory computer science lab assignments to “shortcutting” by generative AI tools, such as ChatGPT. By analyzing the work of three distinct student personas on these assignments, we identified key characteristics of language and structure that influence an assignment's vulnerability to AI abuse. Based on these insights, we propose strategies for educators to adapt labs to both counteract AI shortcutting and encourage productive uses of AI.

The Mathematical And Historical Significance Of The Four-Color Theorem, Brock Bivens

Scholars Day Conference

Computers becoming more readily used in mathematics.

Using Data Mining To Analyze Job Reviews, Nicholas Bornkamp, Tony Breitzman

STEM Student Research Symposium Posters

Job review websites like Glassdoor are not always clear on how well the company operates, especially as viewed from differing levels of employment. For instance, a middle or upper manager from Amazon may have an overall positive review of the company with minor issues about it, but someone who works in the warehouse may have a mixed experience. To solve this issue and determine any correlation between employee level and their review, data mining techniques were utilized such as website scraping and neural network training to develop a model that analyzes employee reviews.

Self-Sovereign Digital Identities, Maryam M. Ahmed, Bijayata Shrestha, Nick Ivanov

STEM Student Research Symposium Posters

In today's digital landscape, the dominance of internet giants over our digital identities raises concerns regarding user control and privacy. These companies aggregate vast amounts of data from diverse sources, ranging from online interactions to personal information shared on their platforms. This centralized control impedes individual autonomy and privacy. To address these challenges, we propose Self-Sovereign Digital Identities (SSDIs) as a solution. SSDIs empower individuals with control over their online identity information, encompassing ownership, security, privacy, and portability. By decentralizing identity management, SSDIs offer users autonomy and enhance privacy protection. Moreover, we introduce Sans-Chain Smart Contracts, a novel approach to …

Utilizing Machine Learning To Predict Workplace Violence In Hospitals, Aiden Touhill, Carter Profico, Avery Bobbitt, Joe Dipietro, Christopher Duym, Anthony Ung, Jack Myers

STEM Student Research Symposium Posters

Random forest machine learning models are a form of classification model, which attempts to sort data into one of two predefined categories. When trained on a set of data from a hospital, where each entry is listed as either conditions for workplace violence or not, a random forest model can begin to classify new data as it comes in. We developed a way to automatically poll hospital systems for the required data needed to make a prediction on the potential for workplace violence at any one given moment. Our team was unable to gain access to real hospital data, so …

Missionlog R - Air Force Mission History Report Management System With Encryption & Database Integration, Matthew Bachrach, Sam Jeffery, Michael Lim, Joe Johnston, Jack Healy, Andrew Siciliano, Jack F. Myers

STEM Student Research Symposium Posters

MissionLog R is a mobile and web application designed to allow for the filing, confirmation, and finalization of a mission history report for the US Air Force. The mobile application is used by filing members to fill out their report on their issued iPad. The form has been fitted with data validation and ghost data so the user knows what type of input that specific field is looking for. The web application can be used by the filing member to import their report from their iPad if no WiFi was available when they filled out the report. Then, the confirming …

React Native Photo & Video Streaming/Processing Api Integration, Anamaria Oharciuc, William Carr, Sushanth Ambati, Ryan Blaisdell, Kyle Reed, Jack F. Myers

STEM Student Research Symposium Posters

RunSignup is a software company specializing in event management technology. Event organizers, known as Race Directors, utilize their platform to create and manage events, organize media albums, stream their events to the public, and more. In order to allow Race Directors to complete these actions in real-time at their events, RunSignup asked our team to develop a mobile app. With the app, Race Directors would be able to upload and stream photos to the event’s photo albums as soon as they are taken, and they could livestream the event to YouTube directly from their mobile device. Even if the device …

Blueberry Drone Ai: Smart Farming Of Blueberries Using Artificial Intelligence And Autonomous Drones, Robert Czarnota, Anthony Segrest, Anthony Thompson, Harper Zappone, Hieu Nguyen, Nguyen Thanh, Ik Jae Lee, Lori Green, Tuan Le

STEM Student Research Symposium Posters

This project seeks to assist blueberry growers in New Jersey with preventing blueberry scorch disease. Plants can’t be cured of scorch, so they have to be removed to prevent the disease from spreading to other bushes. This project aims to use object detection and classifier machine learning models in order to detect scorch disease with photos from intelligent drones. Images are first tiled, then processed through and convolutional neural network that detects scorch symptoms. Lastly, a fully connected neural network is implemented to make a final prediction.

Ai-Powered Learning: Blending Ai With Active Learning In The Information Literacy Classroom, Kevin J. Reagan, Wilhelmina Randtke

Georgia International Conference on Information Literacy

In 2016, the ACRL Framework for Information Literacy in Higher Education launched in response to more voluminous, less-vetted online information, including misinformation and content farms. Subsequently, the ACRL Framework has been widely adopted, and numerous high-quality lesson plans and resources for teaching the frames already exist, including published lesson plans and textbooks. Now, generative AI tools, such as ChatGPT and other chat bots present new challenges for information literacy educators. For instance, in addition to teaching students how to identify issues such as fake news, the information literacy professional has to address topics such as ethical AI use, AI hallucination …

A Design Science Approach To Investigating Decentralized Identity Technology, Janelle Krupicka

Cybersecurity Undergraduate Research Showcase

The internet needs secure forms of identity authentication to function properly, but identity authentication is not a core part of the internet’s architecture. Instead, approaches to identity verification vary, often using centralized stores of identity information that are targets of cyber attacks. Decentralized identity is a secure way to manage identity online that puts users’ identities in their own hands and that has the potential to become a core part of cybersecurity. However, decentralized identity technology is new and continually evolving, which makes implementing this technology in an organizational setting challenging. This paper suggests that, in the future, decentralized identity …

Use Of Deep Learning In Content-Based Image Retrieval (Cbir), Angelina Das

ATU Research Symposium

In the world of computer vision and data retrieval, a crucial task is finding images within a database based on their visual content. This is known as content-based image retrieval (CBIR). As the number of digital images explodes across fields like online shopping, healthcare, and social media, the need for powerful and precise CBIR systems becomes ever more critical. Early CBIR methods depended on features crafted by hand, like color distributions, texture descriptions, and shape characteristics. However, these techniques often have difficulty capturing the true meaning of an image and might not handle very large datasets effectively. With the rise …

Optimizing Keyboard Layouts For English Text, David Sommerfield

Research & Creative Achievement Day

QWERTY has been the de facto layout for English text input since its invention in 1874. Its continued usage has led to concerns about its ergonomic shortcomings. Previous attempts at layout creation have usually relied on manual observations of typing data rather than a predictive model. To address this issue, we propose a methodology that incorporates both corpus data from 22 million English websites and 8,228 hours of real-world typing data from participants. The corpus data is processed into bigrams and their number of occurrences. The typing data is preprocessed to exclude user-made typos, and then each bigram is tabulated …

Binder, Tyler A. Peaster, Lindsey M. Davenport, Madelyn Little, Alex Bales

ATU Research Symposium

Binder is a mobile application that aims to introduce readers to a book recommendation service that appeals to devoted and casual readers. The main goal of Binder is to enrich book selection and reading experience. This project was created in response to deficiencies in the mobile space for book suggestions, library management, and reading personalization. The tools we used to create the project include Visual Studio, .Net Maui Framework, C#, XAML, CSS, MongoDB, NoSQL, Git, GitHub, and Figma. The project’s selection of books were sourced from the Google Books repository. Binder aims to provide an intuitive interface that allows users …

Jsper (Just Stablediffusion Plus Easy Retraining), Adam Rusterholz, Meghan Finn, Zach Zolliecoffer, Zach Judy

ATU Research Symposium

JSPER is an an AI art generation Web Application that is both flexible and accessible. Our goal is to enable anyone to create and use their own customized art models, regardless of technical skill level. These models can be trained on almost anything, from a person, to an animal, to a specific object, or even style. The user only has to upload a handful of images of their subject. Then, training settings get optimized at the push of a button to match the type of subject the user is training. After training, their customized model can be used to generate …

Optimizing Campus Chat-Bot Experience Using Puaa: Integrating Large Language Model (Llm) Into University Ai Assistants, Sijan Panday, Zurab Sabakhtarishvili, Clayton Jensen

ATU Research Symposium

The advent of large language models (LLMs) such as Chat-GPT and Bard marks a significant milestone in knowledge acquisition, offering a streamlined alternative to the traditionally labor-intensive process of navigating through multiple checkpoints on the web. This emerging trend in LLMs renders the prevalent rule-based chatbots, commonly utilized by universities, increasingly outdated and subpar. This research project proposes integrating LLM technology into university websites, specifically targeting the needs of students seeking information about their institutions by introducing PUAA (Personal University AI Assistant). Our approach involves using the Retrieval-Augmented Generation (RAG) framework, leveraging the capabilities of the LlamaIndex in conjunction with …

Genetic Association In Entylia Carinata Using Random Forest Classification, Caden J. Harper

Research & Creative Achievement Day

The goal of this research was to identify locations in the genome of the Entylia carinata, known as the treehopper, that are associated with anomalous behavior exhibited by the species. Treehoppers are phytophagous insects and are shown to feed, reproduce, and rear their young on specific aster species. Observation has shown that the insects will disregard potential mates in close proximity in favor of those that originate from the same plant species as themselves. This behavior suggests genetic separation in the species based on plant nativity and warrants genetic analysis. Machine learning offers an effective genetic association technique due to …

Spoton, Corey A. Naegle, Caleb Mcclure, Chase M. Tallon, Holden J. O'Neal

ATU Research Symposium

SpotOn is a project developed to solve problems with owners losing their pets. The project is in short a solar-powered dog harness with GPS capability with its own application for mobile devices.

Techniques To Detect Fake Profiles On Social Media Using The New Age Algorithms – A Survey, A K M Rubaiyat Reza Habib, Edidiong Elijah Akpan

ATU Research Symposium

This research explores the growing issue of fake accounts in Online Social Networks [OSNs]. While platforms like Twitter, Instagram, and Facebook foster connections, their lax authentication measures have attracted many scammers and cybercriminals. Fake profiles conduct malicious activities, such as phishing, spreading misinformation, and inciting social discord. The consequences range from cyberbullying to deceptive commercial practices. Detecting fake profiles manually is often challenging and causes considerable stress and trust issues for the users. Typically, a social media user scrutinizes various elements like the profile picture, bio, and shared posts to identify fake profiles. These evaluations sometimes lead users to conclude …

Enhancing R2l Intrusion Detection Using Decision Trees, Stephen Sommer

Research & Creative Achievement Day

In the age of advancing technology, artificial intelligence, and big data, Remote to Local (R2L) attacks are increasingly threatening cloud computing environments, heightening concerns about security and privacy. Intrusion detection systems (IDS) using Artificial Intelligence play a role in safeguarding data integrity within databases by swiftly identifying and isolating suspicious records. Furthermore, machine learning techniques enhance the effectiveness of these IDS by continuously adapting to new attack patterns and improving accuracy. This research investigates the use of Decision Tree, a Machine Learning Algorithm for enhancing Remote to Local (R2L) intrusion detection capabilities, utilizing the KDD Cup 1999 dataset and the …

Anomaly Detection With Spiking Neural Networks (Snn), Shruti Bhandari, Vyshnavi Gogineni

ATU Research Symposium

Abstract:

Anomaly detection, the identification of rare or unusual patterns that deviate from normal behavior, is a fundamental task with wide-ranging applications across various domains. Traditional machine learning techniques often struggle to effectively capture the complex temporal dynamics present in real-world data streams. Spiking Neural Networks (SNNs), inspired by the spiking nature of biological neurons, offer a promising approach by inherently modeling temporal information through precise spike timing. In this study, we investigate the use of Spiking Neural Networks (SNNs) for detecting anomalies or unusual patterns in data. We propose an SNN model that can learn what constitutes normal …

Innovating Inventory And Alert Systems With Object Tracking, Juan Harmse, Esther Peden

Campus Research Day

Security system users require safeguarding inventory from potential theft while reducing manual tracking of physical objects. Our contribution harnesses the power of artificial intelligence and computer vision with YOLO to automate the process of tracking inventory items. The system sends alerts to the inventory manager when it detects particular events. Our approach was evaluated with KernProf profiling, interference, and orientation tests. The results were overall positive in these testing areas.