Asymptotically Optimal Agents
نویسندگان
چکیده
Artificial general intelligence aims to create agents capable of learning to solve arbitrary interesting problems. We define two versions of asymptotic optimality and prove that no agent can satisfy the strong version while in some cases, depending on discounting, there does exist a non-computable weak asymptotically optimal agent.
منابع مشابه
Optimistic Agents Are Asymptotically Optimal
We use optimism to introduce generic asymptotically optimal reinforcement learning agents. They achieve, with an arbitrary finite or compact class of environments, asymptotically optimal behavior. Furthermore, in the finite deterministic case we provide finite error bounds.
متن کاملAsymptotically Optimal Deterministic Rendezvous
In this paper, we address the deterministic rendezvous in graphs where k mobile agents, disseminated at different times and different nodes, have to meet in finite time at the same node. The mobile agents are autonomous, oblivious, labeled, and move asynchronously. Moreover, we consider an undirected anonymous connected graph. For this problem, we exhibit some asymptotical time and space lower ...
متن کاملThe linear saturated decentralized strategy for constrained flow control is asymptotically optimal
We present an algorithm for constrained network flow control in the presence of an unknown demand. Our algorithm is decentralized in the sense that it is implemented by a team of agents, each controlling just the flow on a single arc of the network based only on the buffer levels at the nodes at the extremes of the arc, while ignoring the actions of other agents and the network topology. We pro...
متن کاملStability Analysis and Optimal Control of Vaccination and Treatment of a SIR Epidemiological Deterministic Model with Relapse
In this paper, we studied and formulated the relapsed SIR model of a constant size population with standard incidence rate. Also, the optimal control problem with treatment and vaccination as controls, subject to the model is formulated. The analysis carried out on the model, clearly showed that the infection free steady state is globally asymptotically stable if the bas...
متن کاملEconomic Recommendation Systems
In the on-line Explore & Exploit literature, central to Machine Learning, a central planner is faced with a set of alternatives, each yielding some unknown reward. The planner’s goal is to learn the optimal alternative as soon as possible, via experimentation. A typical assumption in this model is that the planner has full control over the experiment design and implementation. When experiments ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2011