Latest posts tagged with #multiaagent on Bluesky
Dual‑Dial Control Boosts Multi‑Agent AI Reasoning Accuracy
MACI adds two dials—an information filter and a behavior controller—to steer debates. In clinical diagnosis and news bias detection tests, it cut token use while preserving accuracy. Read more: getnews.me/dual-dial-control-boosts... #maci #multiaagent
Interactive World Latent Boosts Team Coordination in Multi‑Agent AI
IWoL embeds inter‑agent relations in a shared latent space, improving coordination for multi‑agent reinforcement learning; posted 29 Sept 2025, revised 1 Oct 2025. Read more: getnews.me/interactive-world-latent... #interactivelatent #multiaagent
AutoMaAS: Self‑Evolving Multi‑Agent Search for Large Language Models
AutoMaAS, unveiled Oct 3 2025, improves benchmark scores by up to 7.1% while cutting inference costs 3–5%, automatically evolving multi‑agent LLM pipelines. getnews.me/automaas-self-evolving-m... #automaas #llm #multiaagent
AdvEvo-MARL Boosts Safety in Multi-Agent AI Systems
AdvEvo‑MARL reduces jailbreak attack success to under 20%, versus up to 38.33% for baselines, while task accuracy can improve by up to 3.67% on reasoning benchmarks. Read more: getnews.me/advevo-marl-boosts-safet... #advevomarl #aisafety #multiaagent
Guided Policy Search Boosts Stability for Multi-Agent Dynamic Games
MA‑GPS (Multi-Agent Guided Policy Search) uses regularization with updates to stabilize training, achieving faster convergence in vehicle platooning tasks. Read more: getnews.me/guided-policy-search-boo... #multiaagent #reinforcementlearning
Decentralized Cooperative Multi-Agent RL through Context Modeling
DAC (Dynamics‑Aware Context) models multi‑agent RL as a contextual MDP, using latent contexts to reset values when policies shift and beating Q‑learning on SMAC. Read more: getnews.me/decentralized-cooperativ... #dac #multiaagent
