Hierarchical Reinforcement Learning with Unlimited Recursive Subroutine Calls

机译：无限递归子例程调用的分层强化学习

获取原文

页面导航

摘要
著录项
相似文献
相关主题

摘要

Humans can set suitable subgoals to achieve certain tasks. They can also set sub-subgoals recursively if required. The depth of this recursion is apparently unlimited. Inspired by this behavior, we propose a new hierarchical reinforcement learning architecture called RGoal. RGoal solves the Markov Decision Process (MDP) in an augmented state-action space. In multitask settings, sharing subroutines between tasks makes learning faster. A novel mechanism called thought-mode is a type of model-based reinforcement learning. It combines learned simple tasks to solve unknown complicated tasks rapidly, sometimes in zero-shot time.

机译：人类可以设置合适的子目标来完成某些任务。如果需要，他们还可以递归设置子子目标。递归的深度显然是无限的。受此行为的启发，我们提出了一种称为RGoal的新的分层强化学习架构。 RGoal在增强的状态行动空间中解决了马尔可夫决策过程（MDP）。在多任务设置中，在任务之间共享子例程可以使学习更快。一种称为思维模式的新颖机制是一种基于模型的强化学习。它结合了学到的简单任务，可以快速解决未知的复杂任务，有时甚至可以零触发。

著录项

来源
《International Conference on Artificial Neural Networks》|2019年|103-114|共12页
会议地点
作者
Yuuji Ichisugi; Naoto Takahashi; Hidemoto Nakada; Takashi Sano;
展开▼
作者单位

展开▼
会议组织
原文格式 PDF
正文语种
中图分类
关键词
Hierarchical reinforcement learning; Model-based reinforcement learning; Zero-shot learning; Computational neuroscience;

机译：分层强化学习;基于模型的强化学习;零镜头学习;计算神经科学;

相似文献

外文文献
中文文献
专利

1. Neural signature of hierarchically structured expectations predicts clustering and transfer of rule sets in reinforcement learning [J] . Collins Anne Gabrielle Eva, Frank Michael Joshua Cognition: International Journal of Cognitive Psychology . 2016,第Null期

机译：分层结构的期望的神经签名可预测强化学习中规则集的聚类和转移
2. Hierarchically organized behavior and its neural foundations:A reinforcement learning perspective [J] . Matthew M. Botvinick, Yael Niv, Andrew C. Barto Cognition: International Journal of Cognitive Psychology . 2009,第3期

机译：分层组织的行为及其神经基础：强化学习视角
3. A biologically inspired hierarchical reinforcement learning system [J] . Zhou WD, Coggins R Cybernetics and Systems . 2005,第1期

机译：受生物启发的分层强化学习系统
4. Hierarchical Reinforcement Learning with Unlimited Recursive Subroutine Calls [C] . Yuuji Ichisugi, Naoto Takahashi, Hidemoto Nakada, International Conference on Artificial Neural Networks . 2019

机译：使用无限递归子程序调用的分层加强学习
5. Learning state and action space hierarchies for reinforcement learning using action -dependent partitioning. [D] . Asadi, Mehran. 2006

机译：使用依赖于动作的分区来学习状态和动作空间层次结构，以进行强化学习。
6. Computational evidence for hierarchically structured reinforcement learning in humans [O] . Maria K. Eckstein, Anne G. E. Collins 2020

机译：人类分层结构强化学习的计算证据
7. Hierarchically Structured Reinforcement Learning for Topically Coherent Visual Story Generation [O] . Qiuyuan Huang, Zhe Gan, Asli Celikyilmaz, 2019

机译：用于局部连贯的视觉故事的分层结构强化学习
8. Intrinsically Motivated Reinforcement Learning: A Promising Framework for Developmental Robot Learning [R] . Stout, A. , Konidaris, G. D. , Barto, A. G. 2005

机译：本质动机强化学习：发展机器人学习的有前途的框架

Hierarchical Reinforcement Learning with Unlimited Recursive Subroutine Calls

摘要

著录项

相似文献

相关主题

期刊订阅