Ized Action Space
نویسندگان
چکیده
Recent work has shown that deep neural networks are capable of approximating both value functions and policies in reinforcement learning domains featuring continuous state and action spaces. However, to the best of our knowledge no previous work has succeeded at using deep neural networks in structured (parameterized) continuous action spaces. To fill this gap, this paper focuses on learning within the domain of simulated RoboCup soccer, which features a small set of discrete action types, each of which is parameterized with continuous variables. The best learned agents can score goals more reliably than the 2012 RoboCup champion agent. As such, this paper represents a successful extension of deep reinforcement learning to the class of parameterized action space MDPs.
منابع مشابه
A Review of the Concepts of Social Action and Isolation in Virtual Space
Cyberspace and its impact as the main competitor of real space in various aspects is considered and have been studied by many thinkers and theorists. For various reasons (political, social, cultural, etc.) it is lead to the presence of people, especially young people in virtual space, as all borders crossed the behavior and influence actions of people. According to the increasing importance and...
متن کاملP-subgraph Isomorphism Computation and Upper Bound Complexity Estimation
An approach for subgraph isomorphism computation of parameter-ized graphs will be presented. Parameterized graphs (short: p-graphs) are extensions of undirected graphs by parameter vectors at the nodes and edges. We will deene p-graphs and basic concepts of subgraph isomorphism computation for p-graphs. A bottom-up algorithm for p-subgraph isomorphism computation according to a given search gra...
متن کاملOrbit Spaces Arising from Isometric Actions on Hyperbolic Spaces
Let be a differentiable action of a Lie group on a differentiable manifold and consider the orbit space with the quotient topology. Dimension of is called the cohomogeneity of the action of on . If is a differentiable manifold of cohomogeneity one under the action of a compact and connected Lie group, then the orbit space is homeomorphic to one of the spaces , , or . In this paper we suppo...
متن کاملSolutions to the Communication Minimization Problem for Affine Recurrence Equations
This paper deals with communication optimization which is a crucial issue in automatic parallelization. From a system of parameter-ized aane recurrence equations, we propose a heuristic which determines an eecient space-time transformation. It reduces rst the distant communications and then the local communications.
متن کاملTwo-level Preconditioners for Ill-conditioned Linear Systems with Semideenite Regularization
A family preconditioners for the solution of discrete linear systems arising in regular-ized ill-posed problems is presented. These preconditioners are based on a two-level splitting of the solution space, and were previously developed by Hanke and Vo-gel for positive deenite regularization operators. The work presented here extends previous results to the case where the regularization operator...
متن کامل