About
Research & Development
Unpacking Gradient Inversion Risks in Federated Learning: A Deep Dive into Model Vulnerabilities
TLDR: This research paper investigates the real-world feasibility of Gradient Inversion Attacks (GIAs) in Federated Learning (FL), challenging common assumptions that often overstate attack...
Research & Development
Optimizing Multitask Model Creation with PSO-Merging
TLDR: PSO-Merging is a novel data-driven method for combining multiple expert language models into a single, efficient multitask model. It leverages Particle Swarm Optimization...
Research & Development
Generative AI: A New Frontier in Autonomous Driving System Testing
TLDR: A survey of 91 studies reveals that Generative AI is increasingly vital for testing Autonomous Driving Systems (ADS). It's primarily used for scenario-based...
Research & Development
AI Models Advance Detection of Inappropriate Language in Medical Training Materials
TLDR: A new study introduces AI models to automatically detect inappropriate language (IUL) in medical school curricula, which includes outdated, exclusionary, or non-patient-centered terms....
News & Current Events
Duke University Enhances AI Suite with GPT-5 Integration into DukeGPT
TLDR: Duke University has significantly upgraded its institutional AI platform, DukeGPT, by integrating OpenAI's most advanced model, GPT-5. This enhancement provides users with superior...
Research & Development
A New Framework for Detecting Imaginary Objects in AI-Generated Descriptions
TLDR: GLS IM is a novel, training-free framework that detects object hallucinations in Large Vision-Language Models (LVLMs) by combining global and local similarity signals...