Optimizing Speech Models by Exploiting Token Redundancy
Discover how exploiting redundancy in large speech models boosts efficiency by 27% without losing accuracy using Affinity Pooling.
Physical Adversarial Attacks on AI Surveillance Systems
Explore physical adversarial attacks on AI surveillance, focusing on detection, tracking, and visible-infrared evasion techniques in real-world scenarios.
Uncovering Tone-Based Bias in LLMs and Emoji Models
Explore how tone-based biases in LLMs and emoji embeddings impact digital communication and why addressing these AI biases is crucial for inclusion.
CyberAgent Boosts Efficiency with ChatGPT Enterprise & Codex
Discover how CyberAgent enhances productivity and security by integrating ChatGPT Enterprise and Codex for faster AI-driven innovation.
MedDialBench: Testing LLM Diagnostic Robustness in Medical Dialogue
MedDialBench benchmarks LLM diagnostic accuracy under varied patient behaviors, revealing impacts of symptom fabrication and non-cooperation on model perfo...