XFED: Non-Collusive Model Poisoning Attack on Federated Learning
Discover XFED, a non-collusive model poisoning attack that bypasses defenses in federated learning without attacker coordination.
SafeMind: Adaptive Safe Control for Quadruped Robots
SafeMind offers a risk-aware control framework ensuring adaptive, safe quadruped locomotion with reduced safety violations and energy use.
SafeAdapt: Provably Safe Policy Updates in Deep RL
SafeAdapt ensures provably safe policy updates in deep reinforcement learning, preserving safety during adaptation in dynamic environments.
ECHO: Fast Chest X-ray Report Generation with Diffusion AI
ECHO enables rapid, accurate chest X-ray report generation using one-step block diffusion, boosting speed by 8× and improving report quality significantly.
Many-Tier Instruction Hierarchy for Advanced LLM Agents
Explore Many-Tier Instruction Hierarchy (ManyIH) to improve how LLM agents handle conflicting instructions with scalable privilege levels.