Auditing AI Benchmarks: Stop Reward Hacking with BenchJack
Discover how BenchJack audits AI agent benchmarks to prevent reward hacking and improve evaluation accuracy for reliable AI performance.
MAVIC: Macro-Action Value Correction for Multi-Agent Instruction Compliance
Discover MAVIC, a novel method improving multi-agent reinforcement learning by correcting value estimates for better instruction compliance and task perfor...
Verifier-Guided Action Selection Boosts Embodied Agents
Enhance embodied AI agents' decision-making with Verifier-Guided Action Selection, improving robustness and performance in complex tasks.
FQPDR: Quantum Federated Learning for Early Diabetic Retinopathy Detection
Discover FQPDR, a federated quantum neural network enhancing privacy-preserving early detection of diabetic retinopathy with efficient, secure AI models.
Reciprocity Gradient: Boosting AI Strategic Cooperation
Discover how the Reciprocity Gradient optimizes AI agents' strategic interactions by enhancing cooperation and reputation management in multi-agent systems...