12 Matching Results

Search Results

Results: 1 - 12
open access

A Systematic Review of Large Language Models in Medical Specialties: Applications, Challenges and Future Directions

Description: Article synthesizes evidence from 84 studies published between January 2021 and March 2024 on the use of large language models (LLMs) across medical specialties, highlighting applications in clinical NLP, diagnostic support, and medical education. Although LLMs demonstrate high accuracy in certain narrowly defined tasks, their performance remains highly variable and methodologically inconsistent, underscoring the need for domain‑specific models and rigorous validation standards to ensure safe c… more
Date: June 11, 2025
Creator: Alkalbani, Asma Musabah; Alrawahi, Ahmed Salim; Salah, Ahmad; Haghighi, Venus; Zhang, Yang; Alkindi, Salam et al.
Item Type: Refine your search to only Article
Partner: UNT College of Information
open access

Optimizing Intrusion Detection in IoMT Networks Through Interpretable and Cost-Aware Machine Learning

Description: Article presents a cybersecurity framework for Internet of Medical Things (IoMT) networks using a fine‑tuned XGBoost classifier, achieving high attack‑detection accuracy while maintaining interpretability through SHAP‑based feature analysis. Comparative evaluation with Logistic Regression and a late‑fusion max‑voting model shows that the fused approach offers improved precision and reduced false negatives, providing a balanced, cost‑efficient, and robust solution for securing modern IoMT enviro… more
Date: May 9, 2025
Creator: Hafid, Abdelatif & Rahouti, Mohamed
Item Type: Refine your search to only Article
Partner: UNT College of Information
open access

A Blockchain-Assisted Federated Learning Framework for Secure and Self-Optimizing Digital Twins in Industrial IoT

Description: Article proposes a secure and adaptive framework for optimizing digital twins in the Industrial Internet of Things by integrating blockchain, federated learning, and explainable AI to enhance trust, privacy, and model accuracy. Article demonstrates through real‑world evaluation that this combined approach strengthens security, transparency, and interpretability while improving the efficiency and reliability of self‑optimizing digital‑twin systems.
Date: January 2, 2025
Creator: Ababio, Innocent Boakye; Bieniek, Jan; Rahouti, Mohamed; Hayajneh, Thaier; Aledhari, Mohammed; Verma, Dinesh C. et al.
Item Type: Refine your search to only Article
Partner: UNT College of Information
open access

REU Site: Knowledge Beyond Language with Vector Embeddings

Description: Data management plan for the grant, "REU Site: Knowledge Beyond Language with Vector Embeddings." This three-year REU program at the University of North Texas will train undergraduate students to develop and apply AI systems that use reusable vector embeddings to share knowledge across domains. Through interdisciplinary research projects, students will learn to create, evaluate, and contribute transferable AI representations that improve machine learning applications while gaining hands-on rese… more
Date: 2026-10-01/2029-09-30
Creator: Xiao, Ting
Item Type: Refine your search to only Text
Partner: UNT College of Information

Metadata Quality Assurance for ETDs Collections

Description: Presentation covers practical aspects of assuring quality of Electronic Theses and Dissertations (ETDs) collections metadata, including reviewing, quality control, and correcting inconsistencies. It was presented at the UNT Libraries' 2026 Student Snapshots Symposium held in Denton, Texas.
Date: April 8, 2026
Creator: Shrivastava, Tushar
Item Type: Refine your search to only Presentation
Partner: UNT Libraries
open access

Robust Federated Learning Against Data Poisoning Attacks: Prevention and Detection of Attacked Nodes

Description: Article presents a robust federated learning defense framework that combines a prevention strategy, Confident Federated Learning, which identifies and removes mislabeled local samples, with a novel detection method that flags malicious workers through class‑wise clustering of neuron activation patterns. Together, these approaches effectively mitigate data‑poisoning attacks, ensure rapid global‑model convergence, and maintain resilience across data distributions and model sizes without requiring… more
Date: July 24, 2025
Creator: Ovi, Pretom Roy & Gangopadhyay, Aryya
Item Type: Refine your search to only Article
Partner: UNT College of Information
open access

Content-based quality evaluation of scientific papers using coarse feature and knowledge entity network

Description: Article introduces a content‑based, interpretable framework for pre‑evaluating scientific paper quality by defining key attributes - integrity, clarity, novelty, and significance - and modeling them through fine‑grained metadata and knowledge‑network features. Article demonstrates that this approach, validated on the ICLR dataset with Random Forest achieving F1 scores of 0.715 and 0.762, enhances machine‑learning performance and provides transparent, feature‑driven evidence for assessing scient… more
Date: July 9, 2024
Creator: Wang, Zhongyi; Zhang, Haoxuan; Chen, Haihua; Feng, Yunhe & Ding, Junhua
Item Type: Refine your search to only Article
Partner: University of North Texas
Back to Top of Screen