Physical Sciences and Mathematics Commons^™

Image Captioning Using Reinforcement Learning, Venkat Teja Golamaru

Master's Projects

Image captioning is a crucial technology with numerous applications, including enhancing accessibility for the visually impaired, developing automated image indexing and retrieval systems, and enriching social media experiences. However, accurately describing the content of an image in natural language remains a challenge, particularly in low-resource settings where data and computational power are limited. The most advanced image captioning architectures currently use encoder-decoder structures that incorporate a sequential recurrent prediction model. This study adopts a typical Convolutional Neural Network (CNN) encoder Recurrent Neural Network (RNN) decoder structure for image captioning, but it has framed the problem as a sequential decision-making task. …

Influence Maximization Based On Community Detection And Dominating Sets, Ameya Marathe

Master's Projects

An online platform where various people come together to share information and communicate is called a social network. These platforms are set apart from other means of communication mostly because you can follow and interact also with different people even some you never met, comment on their posts, and re-sharing their posts. Companies such as Amazon and Walmart use these platforms daily for marketing purposes, like spreading information regarding new products and services they offer. They carefully select a subset of users, called influencers, who are usually the ones with high influence over the rest of the users. Influencers receive …

Relationalnet Using Graph Neural Networks For Social Recommendations, Dharahas Tallapally

Master's Projects

Traditional recommender systems create models that can predict user interests based on the user-item relationships. However, these systems often have limited performance due to sparse user behavior data. To address this challenge, researchers are now exploring models for social recommendation that can account for both user- user and user-item relationships based on social networks, and user past behavior, respectively. These models aim to understand each user’s behavior by considering their trusted neighbors and their influence on each other. Specifically, the potential embedding of each user is influenced by their trusted neighbors, who are, in turn, influenced by their own trusted …

Container Caching Optimization Based On Explainable Deep Reinforcement Learning, Divyashree Jayaram

Master's Projects

Serverless edge computing environments use lightweight containers to run IoT services on a need basis i.e only when a service is requested. These containers experience a cold start up latency when starting up. One probable solution to reduce the startup delay is container caching on the edge nodes. Edge nodes are nodes that are closer in proximity to the IoT devices. Efficient container caching strategies are required since the resource availability on these edge devices is limited. Because of this constraint on resources, the container caching strategies should also take proper resource utilization into account. This project tries to further …

Resource Coordination Learning For End-To-End Network Slicing Under Limited State Visibility, Xiang Liu

Master's Projects

This paper discusses a resource coordination problem under limited state visibility to realize end-to-end network slices that are hosted by multiple network domains. We formulate this resource coordination problem as a special type of the multi- armed bandit (MAB) problem called the combinatorial multi-armed bandit (CMAB) problem. Based on this formulation, we convert the problem to a regret minimization problem with a linear objective function and solve it by adapting the Learning with Linear Rewards (LLR) algorithm. In this paper, we present a new hybrid approach that incorporates state reports, which include partial resource information in each domain, into the …

The Bias Report: An Automated News Aggregator For Political Bias Classification And News Summarization, Anant Joshi

Master's Projects

Political polarization is on the rise in the US, driven in large part by divisive news that goes viral on the Internet. Specifically, many media outlets use slanted language and publish misinformation in order to drive user traffic and engagement. Almost 80% of US citizens get their news from online sources, but there is a lack of public safeguards against biased news. A large amount of news is published online every day by media organizations, and it is impossible to manually analyze this amount of data. There is a clear need for automated, public-facing solutions in the current political climate …

Multi-Label Text Classification With Transfer Learning, Likhitha Yelamanchili

Master's Projects

Multi-label text categorization is a crucial task in Natural Language Processing, where each text instance can be simultaneously assigned to numerous labels. This project's goal is to assess how well several deep learning models perform on a real-world dataset for multi-label text classification. We employed data augmentation techniques like Synonym Substitution and Random Word Substitution to address the problem of data imbalance. We conducted experiments on a toxic comment classification dataset to evaluate the effectiveness of several deep learning models including Bi-LSTM, GRU, and Bi-GRU, as well as fine- tuned pre-trained BERT models. Many metrics, including log loss, recall@k, and …

Sign Language Recognition Using A Hybrid Machine Learning Model, Peeyusha Shivayogi

Master's Projects

Sign Language is a visual language used by millions of people around the world. American Sign Language (ASL) is one of the most popular sign languages and the third most popular language in the United States. Automatic recognition of ASL signs can help bridge the communication gap between deaf and hearing individuals. In this project, we explore the use of deep learning models for ASL sign recognition, using the MNIST dataset as a benchmark. We preprocessed the data by reshaping the images to the input layer size of the models and normalized the pixel values. We evaluated five popular deep-learning …

Analyzing Improvement Of Mask R-Cnn On Arms Plates (And Sponges And Coral), James Lee

Master's Projects

Coral Reefs and their diverse array of life forms play a vital role in maintaining the health of our planet's environment. However, due to their fragility, it can be challenging to study the reefs without damaging their delicate ecosystem. To address this issue, researchers have employed non-invasive methods such as using Autonomous Reef Monitoring Structures (ARMS) plates to monitor biodiversity. Data was collected as genetic samples from the plates, and high-resolution photographs were taken. To make the best use of this image data, scientists have turned to machine learning and computer vision. Prior to this study, MASKR-CNN was utilized as …

Bias Detector Tool For Face Datasets Using Image Recognition, Jatin Vamshi Battu

Master's Projects

Computer Vision has been quickly transforming the way we live and work. One of its sub- domains, i.e., Facial Recognition has also been advancing at a rapid pace. However, the development of machine learning models that power these systems has been marred by social biases, which open the door to various societal issues. The objective of this project is to address these issues and ensure that computer vision systems are unbiased and fair to all individuals. To achieve this, we have created a web tool that uses three image classifiers (implemented using CNNs) to classify images into categories based on …

Video Sign Language Recognition Using Pose Extraction And Deep Learning Models, Shayla Luong

Master's Projects

Sign language recognition (SLR) has long been a studied subject and research field within the Computer Vision domain. Appearance-based and pose-based approaches are two ways to tackle SLR tasks. Various models from traditional to current state-of-the-art including HOG-based features, Convolutional Neural Network, Recurrent Neural Network, Transformer, and Graph Convolutional Network have been utilized to tackle the area of SLR. While classifying alphabet letters in sign language has shown high accuracy rates, recognizing words presents its set of difficulties including the large vocabulary size, the subtleties in body motions and hand orientations, and regional dialects and variations. The emergence of deep …

Navigating Classic Atari Games With Deep Learning, Ayan Abhiranya Singh

Master's Projects

Games for the Atari 2600 console provide great environments for testing reinforcement learning algorithms. In reinforcement learning algorithms, an agent typically learns about its environment via the delivery of periodic rewards. Deep Q-Learning, a variant of Q-Learning, utilizes neural networks which train a Q-function to predict the highest future reward given an input state and action. Deep Q-learning has shown great results in training agents to play Atari 2600 games like Space Invaders and Breakout. However, Deep Q-Learning has historically struggled with learning how to play games with greater emphasis on exploration and delayed rewards, like Ms. PacMan. In this …

3d Ar Reconstruction, Sneh Arvind Kothari

Master's Projects

The goal of the project is to improve the shopping experience for users by using augmented reality technology. People generally want opinions from others when buying shoes offline. Clicking and sending images of a shoe is not an ideal solution as it does not give the complete feel of the shoe. We developed the 3D AR Reconstruction app to make this process better. A user of our app clicks photos of the shoe. This image data is converted to form a mesh that can be shared. On receiving a model the user can open it in the app and interact …

Static Taint Analysis Via Type-Checking In Typescript, Abhijn Chadalawada

Master's Projects

With the widespread use of web applications across the globe, and the ad- vancements in web technologies in recent years, these applications have grown more ubiquitous and sophisticated than ever before. Modern web applications face the constant threat of numerous web security risks given their presence on the internet and the massive influx of data from external sources. This paper presents a novel method for analyzing taint through type-checking and applies it to web applications in the context of preventing online security threats. The taint analysis technique is implemented in TypeScript using its built-in type-checking features, and then integrated into …

Detecting Botnets Using Hidden Markov Model, Profile Hidden Markov Model And Network Flow Analysis, Rucha Mannikar

Master's Projects

Botnet is a network of infected computer systems called bots managed remotely by an attacker using bot controllers. Using distributed systems, botnets can be used for large-scale cyber attacks to execute unauthorized actions on the targeted system like phishing, distributed denial of service (DDoS), data theft, and crashing of servers. Common internet protocols used by normal systems for regular communication like hypertext transfer (HTTP) and internet relay chat (IRC) are also used by botnets. Thus, distinguishing botnet activity from normal activity can be challenging. To address this issue, this project proposes an approach to detect botnets using peculiar traits in …

Ml-Based User Authentication Through Mouse Dynamics, Sai Kiran Davuluri

Master's Projects

Increasing reliance on digital services and the limitations of traditional authentication methods have necessitated the development of more advanced and secure user authentication methods. For user authentication and intrusion detection, mouse dynamics, a form of behavioral biometrics, offers a promising and non-invasive method. This paper presents a comprehensive study on ML-Based User Authentication Through Mouse Dynamics.

This project proposes a novel framework integrating sophisticated techniques such as embeddings extraction using Transformer models with cutting-edge machine learning algorithms such as Recurrent Neural Networks (RNN). The project aims to accurately identify users based on their distinct mouse behavior and detect unauthorized access …

Steganographic Capacity Of Selected Machine Learning And Deep Learning Models, Lei Zhang

Master's Projects

As machine learning and deep learning models become ubiquitous, it is inevitable that there will be attempts to exploit such models in various attack scenarios. For example, in a steganographic based attack, information would be hidden in a learning model, which might then be used to gain unauthorized access to a computer, or for other malicious purposes. In this research, we determine the steganographic capacity of various classic machine learning and deep learning models. Specifically, we determine the number of low-order bits of the trained parameters of a given model that can be altered without significantly affecting the performance of …

Classifying World War Ii Era Ciphers With Machine Learning, Brooke Dalton

Master's Projects

We examine whether machine learning and deep learning techniques can classify World War II era ciphers when only ciphertext is provided. Among the ciphers considered are Enigma, M-209, Sigaba, Purple, and Typex. For our machine learning models, we test a variety of features including the raw ciphertext letter sequence, histograms, and n-grams. The classification is approached in two scenarios. The first scenario considers fixed plaintext encrypted with fixed keys and the second scenario considers random plaintext encrypted with fixed keys. The results show that histograms are the best feature and classic machine learning methods are more appropriate for this kind …

Concept Drift Detection In Android Malware, Inderpreet Singh

Master's Projects

Machine learning and deep learning algorithms have been successfully applied to the problems of malware detection, classification, and analysis. However, most of such studies have been limited to applying learning algorithms to a static snapshot of malware, which fails to account for concept drift, that is, the non-stationary nature of the data. In practice, models need to be updated whenever a sufficient level of concept drift has occurred. In this research, we consider concept drift detection in the context of Android malware. We train a series of Support Vector Machines (SVM) over sliding windows of time and compare the resulting …

Leveraging Tweets For Rapid Disaster Response Using Bert-Bilstm-Cnn Model, Satya Pranavi Manthena

Master's Projects

Digital networking sites such as Twitter give a global platform for users to discuss and express their own experiences with others. People frequently use social media to share their daily experiences, local news, and activities with others. Many rescue services and agencies frequently monitor this sort of data to identify crises and limit the danger of loss of life. During a natural catastrophe, many tweets are made in reference to the tragedy, making it a hot topic on Twitter. Tweets containing natural disaster phrases but do not discuss the event itself are not informational and should be labeled as non-disaster …

Keystroke Dynamics And User Identification, Atharva Sharma

Master's Projects

We consider the potential of keystroke dynamics for user identification and authentication. We work with a fixed-text dataset, and focus on clustering users based on the difficulty of distinguishing their typing characteristics. After obtaining a confusion matrix, we cluster users into different levels of classification difficulty based on their typing patterns. Our goal is to create meaningful clusters that enable us to apply appropriate authentication methods to specific user clusters, resulting in an optimized balance between security and efficiency. We use a novel feature engineering method that generates image-like features from keystrokes and employ multiclass Convolutional Neural Networks (CNNs) to …

Real Time Panoramic Image Processing, Matthew Gerlits

Master's Projects

Image stitching algorithms are able to join sets of images together and provide a wider field of a vision when compared with an image from a single standard camera. Traditional techniques for accomplishing this are able to adequately produce a stitch for a static set of images, but suffer when differing lighting conditions exist between the two images. Additionally, traditional techniques suffer from processing times that are too slow for real time use cases. We propose a solution which resolves the issues encountered by traditional image stitching techniques. To resolve the issues with lighting difference, two blending schemes have been …

Web Traffic Time Series Forecasting, Summanth Redde Mulkkalla

Master's Projects

Online web traffic forecasting is one of the most crucial elements of maintaining and improving websites and digital platforms. Traffic patterns usually predict future online traffic, including page views, unique visitors, session duration, and bounce rates. However, it is challenging to forecast non-stationary online web traffic, particularly when the data has spikes or irregular patterns. This non-stationary property demands a more advanced forecasting technique. In this study, we provide a neural networkbased method, Spiking Neural Networks (SNNs), for dealing with the data spikes and irregular patterns in non-stationary data. In our study, we compared the forecasting results of SNNs with …

Forest Bathing Increases Adolescent Mental Well-Being And Connection To Nature: A Transformative Mixed Methods Study, Jennifer Keller

Antioch University Dissertations & Theses

Previous research has demonstrated that practicing forest bathing has significant positive effects on well-being. However, few studies have investigated whether forest bathing increases adolescent well-being despite the growing adolescent mental health crisis in the United States. Similarly, few studies have explored forest bathing’s impacts on connectedness to nature. Considering the ongoing environmental crisis, determining if forest bathing increases connectedness to nature is a critical expansion of forest bathing research, as connectedness to nature is linked to environmental care and concern. This study investigated the possibility that forest bathing, a nature-based mindfulness practice, could increase adolescent mental well-being and connectedness to …

Children Tell Landscape-Lore Among Perceptions Of Place: Relating Ecocultural Digital Stories In A Conscientizing/Decolonizing Exploration, Meredith Jean Bird Miller

Antioch University Dissertations & Theses

We know that when children feel a sense-of-relation within local natural environments, they are more prone to feel concern for them, while nurturing well-being and resilience in themselves and in lands/waters they inhabit. Positive environmental behaviors often follow into adulthood. Our human capacities for creating sustainable solutions in response to growing repercussions of global warming and climate change may grow if more children feel a sense of belonging in the wild natural world. As educators, if we listen to and learn from students’ voices about how they engage in nature, we can create pedagogical experiences directly relevant to their lives. …

Diving To New Depths: An Exploration Of Aquarium Visitors' Reflection At A Shark Exhibit, Nicole Leigh Conklin

Antioch University Dissertations & Theses

Zoos and aquariums (Z/As) are conservation-oriented free-choice learning institutions. In order to support their mission of advancing wildlife conservation, Z/As deliberately design opportunities and experiences to meaningfully engage visitors in understanding, caring for, and acting on behalf of exhibited species. Conservation psychologists and practitioners have applied values-based and models of human behavior to design and evaluate experiences aimed to influence myriad cognitive, affective, and behavioral outcomes. However, there is little research exploring the role of and opportunity for reflection within these institutions. Models of reflection and reflective practice, which are rooted in both theory and empirical data, stress the importance …

Connecting Antibiotic Resistance To The Environment (Care): Introducing A Novel Framework Integrating Chemical Cross-Resistance And Place-Based Engagement To The Blue Marsh Watershed In Reading, Pennsylvania, Jill Felker

Antioch University Dissertations & Theses

Antibiotic resistance is a serious health threat around the world. Millions of individuals are infected with antibiotic-resistant bacteria yearly, and thousands die from previously curable illnesses. Although antibiotic resistance occurs naturally, misuse of antibiotics accelerates the loss of their effectiveness. Public health campaigns focusing on antibiotic awareness have not effectively communicated and educated the public on this health crisis. New efforts to combat antibiotic resistance are urgently needed. This dissertation focuses on the ecological and public health components of antibiotic resistance research that must be addressed to decelerate antibiotic resistance. A new interdisciplinary theoretical framework was developed to Connect Antibiotic …

Eating Change: A Critical Autoethnography Of Community Gardening And Social Identity, Jessica Gerrior

Antioch University Dissertations & Theses

Community gardening efforts often carry a social purpose, such as building climate resilience, alleviating hunger, or promoting food justice. Meanwhile, the identities and motivations of community gardeners reflect both personal stories and broader social narratives. The involvement of universities in community gardening projects introduces an additional dimension of power and privilege that is underexplored in scholarly literature. This research uses critical autoethnography to explore the relationship of community gardening and social identity. Guided by Chang (2008) and Anderson and Glass-Coffin (2013), a systematic, reflexive process of meaning-making was used to compose three autoethnographic accounts. Each autoethnography draws on the author’s …

Examining The Effects Of Seed Mix Diversity And Composition, Biochar Application, Seeding Rate, Species Identity, And Topography On Palouse Prairie Restoration, Thurman Johnson

EWU Masters Thesis Collection

With over 99.9% of the Palouse prairie lost to land conversion, restoring native plant communities is crucial for ecological function, however, research on Palouse prairie restoration methods is sparse. Seed-based restoration uses a mix of seeded species to enhance competition against weeds, diversify vegetation, and adapt to environmental conditions. However, many factors can be varied, such as seed mix diversity and composition, the proportion of forbs to grasses, and seeding rate, and the most effective levels of each are not clear. Further, soil amendments, such as biochar, may benefit properties of tilled soils, but have not explored in Palouse Prairie …

Investigation Of Small Mammal Species Richness, Abundance, And Genetic Population Structure On And Around The Eastern Washington University Prairie Restoration Site, Sarah Deshazer

EWU Masters Thesis Collection

Small mammals are an ecologically important component of every landscape on Earth. They are a food source for higher trophic level animals, disperse plant seed and mycorrhizal fungi spore, engineer the landscape through burrowing and foraging activities, and alter plant community composition through selective predation of seed and grain. Studies have shown that small mammals may help facilitate the transition between successive stages in prairie restoration. Eastern Washington University has dedicated 120 acres of campus land to restoration of native prairie habitat. Small mammals can play both a positive and a negative role in restoration, therefore it is important to …