analiza aiqbench
This research presents AIQI, the first model-free agent proven to be asymptotically optimal in general reinforcement learning without relying on explicit environment models. By performing universal induction over distributional action-value functions, it expands the theoretical landscape of universal agents and validates optimality for Self-AIXI under a grain-of-truth condition. These findings advance foundational machine learning theory by removing ad-hoc assumptions previously required to prove agent performance in general settings.
A Model-Free Universal AI
arXiv:2602.23242v3 Announce Type: replace Abstract: In general reinforcement learning, all established optimal agents, including AIXI, are model-based, explicitly maintaining and using environment models. This paper introduces Universal AI with Q-Induction (AIQI), the first model-free agent proven to be asymptotically $\varepsilon$-optimal in general RL. AIQI performs universal induction over distributional action-value functions, instead of policies or environments like previous works. Under a grain of truth condition, we prove that AIQI is strong asymptotically $\varepsilon$-optimal and asymptotically $\varepsilon$-Bayes-optimal. We also apply our novel proof techniques to show asymptotic $\varepsilon$-optimality of Self-AIXI without any ad-hoc assumptions. Our results significantly expand the diversity of known universal agents.
Abstract:In general reinforcement learning, all established optimal agents, including AIXI, are model-based, explicitly maintaining and using environment models. This paper introduces Universal AI with Q-Induction (AIQI), the first model-free agent proven to be asymptotically $\varepsilon$-optimal in general RL. AIQI performs universal induction over distributional action-value functions, instead of policies or environments like previous works. Under a grain of truth condition, we prove that AIQI is strong asymptotically $\varepsilon$-optimal and asymptotically $\varepsilon$-Bayes-optimal. We also apply our novel proof techniques to show asymptotic $\varepsilon$-optimality of Self-AIXI without any ad-hoc assumptions. Our results significantly expand the diversity of known universal agents.
Artificial Intelligence (cs.AI)
arXiv:2602.23242 [cs.AI]
(or arXiv:2602.23242v3 [cs.AI] for this version)
Text integral sub licența deschisă a publicației · 109 cuvinte