Planning with an Adaptive World Model

Planning with an Adaptive World Model
复制标题

使用自适应世界模型进行规划

DOI:
--
复制
发表时间:
1990
期刊:
Neural Information Processing Systems
影响因子:
--
通讯作者:
A. Linden
A. Linden
中科院分区:
--
文献类型:
--
作者:
S. Thrun;K. Möller;A. Linden

文献摘要

被引文献

相似文献

我们提出了一种新的连接主义规划方法[TML90]。通过与未知环境的交互,使用梯度下降逐步构建世界模型。为了推导出针对未来增援的最佳行动,规划分为两个步骤:经验网络提出一个计划,随后使用一系列世界模型通过梯度下降进行优化,以便在实际运行时获得最佳增援。机器人应用程序和极平衡任务证明了这种方法的适当性。
We present a new connectionist planning method [TML90]. By interaction with an unknown environment, a world model is progressively constructed using gradient descent. For deriving optimal actions with respect to future reinforcement, planning is applied in two steps: an experience network proposes a plan which is subsequently optimized by gradient descent with a chain of world models, so that an optimal reinforcement may be obtained when it is actually run. The appropriateness of this method is demonstrated by a robotics application and a pole balancing task.