University of Southern Mississippi
Aquila Digital Community (University of Southern Mississippi, USM)Not a member yet
11361 research outputs found
Sort by
2025 Nominee Bookmarks: 3rd-5th Grade
No full texthttps://aquila.usm.edu/magnolia_book_awards/1007/thumbnail.jp
A Comprehensive Performance Comparison of Machine Learning and Federated Learning for Intrusion Detection in Vehicular Ad-Hoc Networks Using CAN-Bus Data
Get PDFFederated Learning (FL) is a Machine Learning (ML) approach that decentralizes training across distributed devices, eliminating the need to centralize data. Unlike traditional ML, where models are trained on aggregated data, FL sends a global model to multiple nodes for local training, with updated parameters transmitted back to the server for aggregation. This process preserves data privacy, making FL ideal for sensitive applications like cybersecurity. However, FL introduces challenges such as data heterogeneity, communication overhead, and difficulties in achieving model convergence, which can impact performance.
This study investigates a fundamental assumption in ML and FL research: that the superior performance of a model in a centralized setting will extend to an FL setup. Specifically, this study explores whether models that excel in centralized ML—such as Neural Networks (NNs) known for handling unstructured data—perform equally well in FL environments where data is split across nodes, each with a limited dataset. This constraint can affect the model\u27s ability to capture intricate patterns, potentially leading to suboptimal performance when aggregating parameters. Given the critical role of cybersecurity, particularly with the rise of autonomous and connected vehicles, it is essential to understand how FL can preserve privacy without affecting the model performance.
The Car Hacking: Attack & Defense Challenge 2020 Dataset, featuring CAN bus traffic data from a Hyundai Avante CN7 with normal and attack messages, was used to simulate a real-world cybersecurity environment. This study focuses on binary classification to evaluate FL models\u27 effectiveness in detecting attacks, highlighting challenges in distributed data environments. Preprocessing involved label encoding and limiting the task to binary detection.
The results reveal that strong centralized ML performance does not always translate to FL. While Naive Bayes excelled in centralized settings, XGBoost performed better in FL, highlighting the need for tailored model selection in distributed environments with limited local training.
These findings underscore the need to tailor FL models to distributed systems\u27 constraints. Future research should examine the effects of more clients and training rounds to refine ML-FL performance dynamics. This study offers insights into developing FL-based IDS, enhancing privacy and adaptability in cybersecurity for connected and autonomous vehicles
Sensory Programming at Your Library: Creating Meaningful, Hands-On Learning Experiences Through Imaginative (and Messy) Play
No full text
Taking Control: Acquiring, Migrating, and Assessing the University\u27s Institutional Repository
No full textIn 2024, the University of Nevada, Reno Libraries assumed ownership of the University’s Institutional Repository, previously managed by a separate campus entity. This transition involved not only acquiring the repository, but also migrating the CMS and executing a fullscale rebrand. This lightning talk will share the Libraries’ experience navigating the acquisition, migration, and assessment of a repository previously managed outside of academic librarianship
The Long and Short of It: IR Considerations for Today and Tomorrow
No full textIn fulfilling a charge to evaluate IR options for University of Arkansas - Little Rock, the library’s IR Task Force had to plan for both short-term considerations (platform selection and implementation) and long-term considerations (sustainable inclusions to meet researcher and institutional needs). One necessary tool developed for implementation was a decision rubric, which identified crucial elements for a successful platform selection and basic functionality. Long-term sustainable practices suggested supporting researcher needs, such as data set hosting, citation metrics, and DOI minting. Attendees will learn how to create a decision rubric and the importance of data set hosting and DOI minting in reference to IR selection, rule compliance, and citation tracking
Cache-Augmented Generation in RAG Pipelines: Fast and Memory-Efficient Approach to Multi-Agent Knowledge Query Systems
Get PDFThis thesis presents an implementation and evaluation of Cache-Augmented Generation (CAG) for knowledge query systems, building upon the approach introduced by Chan et al. (2024). Traditional Retrieval-Augmented Generation (RAG) systems (Lewis et al., 2020) face challenges including high latency, excessive memory usage, and complex infrastructure requirements. By implementing a cache-augmented architecture that preloads relevant knowledge and eliminates real-time retrieval, our approach significantly improves response time while reducing resource requirements. The research demonstrates the effectiveness of CAG through a comprehensive implementation for The University of Southern Mississippi\u27s chatbot system, achieving a 49.02% improvement in response time compared to traditional RAG approaches. Our findings suggest that CAG provides a streamlined and efficient alternative to conventional RAG pipelines, particularly for applications with constrained knowledge bases. This research contributes to the growing field of efficient large language model deployment in resource-conscious environments and offers practical insights for implementing multi-agent knowledge systems in educational settings
Determining the Influence of a Low-Head Dam on Macroinvertebrate Communities in the Bouie River
Get PDFThe Bouie River is a tributary of the Leaf River and belongs to the Pascagoula River watershed, the latter of which has no flow impediments in the form of sills or dams. The Bouie River has been heavily altered in the lower 10 river kilometers (rkm) from aggregate mining (creating a series of deep gravel pits) and the presence of an earthen and concrete low-head dam (sill) located 6 rkm from the mouth of the river. However, despite these alterations, the immediate stretch of river below the low-head dam is a known spawning habitat for diadromous fish species, including the protected Gulf Sturgeon. The purpose of this study was to determine the impact of the low-head dam on aquatic macroinvertebrates communities in the Bouie River to infer ecosystem health by comparing the abundance of Ephemeroptera, Plecoptera, and Trichoptera above and below the structure. From April to June of 2021, 2022, and 2023, a sequence of drift nets was set in tandem in randomized river reaches on either side of the structure. Net contents were preserved in formalin and all macroinvertebrates were sorted and identified to the lowest taxonomic level possible. Macroinvertebrates from fourteen samples were identified with seven samples from below and seven samples from above. I performed an ANOVA test to examine if there were differences in taxa relative abundance (as catch per unit effort) above and below the low-head dam as well as across multiple years. To determine if differences existed in macroinvertebrate community structure, I used non-metric multidimensional scaling plots. I found no difference in abundance of Ephemeroptera, Plecoptera, and Trichoptera above and below the low-head dam nor any differences in community assemblage. My results suggest the low-head dam has minimal influence on the macroinvertebrates in this stretch of the Bouie River, perhaps related to the taxonomic resolution or because of the modified river channel upstream of my above low-head dam sampling locations. Low-head dams generally alter sedimentation and deposition of detritus above such structures; however, the aggregate mined pits likely act as depositional habitats. My data provides baseline information on the macroinvertebrate communities of this tributary, which may aid in restoration success criteria if the structure is removed and may support conservation measures for species of concern in this region
A “Little” Solution to a Big Problem: A Documentary About the Hattiesburg Pocket Museum
Get PDFA “Little” Solution to a Big Problem is a documentary film that revolves around the Hattiesburg Pocket Museum and how it positively impacted its community during the COVID-19 pandemic. The Pocket Museum is a small alleyway located in downtown Hattiesburg, Mississippi and is filled with small objects, scenes, and fun activities. The thesis includes interviews with Vicki Taylor, creator of the Pocket Museum, and Lissa Ortego, the first artist to paint in the alleyway. They explain the history behind the museum, the circumstances that led to its creation, and how they kept the museum relevant during and after the COVID-19 pandemic. This thesis discusses the methods of production and details the reasoning behind many aspects, such as interview questions, lighting selections, and post-production editing decisions. It also discusses why the filmmaker chose a creative route for this thesis and her method for deciding on an expository style of documentary. Within this thesis, it is explained why this documentary was produced and why Vicki’s story was chosen to be cast into the spotlight