基于多智能体的Option自动生成算法

目前分层强化学习中的任务自动分层都是采用基于单智能体的串行学习算法,为解决串行算法学习速度较慢的问题,以Sutton的Option分层强化学习方法为基础框架,提出了一种基于多智能体的Option自动生成算法,该算法由多智能体合作对状态空间进行并行探测并集中应用aiNet实现免疫聚类产生状态子空间,然后并行学习生成各子空间上的内部策略,最终生成Option. 以二维有障碍栅格空间内2点间最短路径规划为任务背景给出了算法并进行了仿真实验和分析.结果表明,基于多智能体的Option自动生成算法速度明显快于基于单智能体的算法.     

分层强化学习中的动态分层方法研究

分层强化学习中现有的自动分层方法均是在对状态空间进行一定程度探测之后一次性生成层次结构,不充分探测不能保证求解质量,过度探测则影响学习速度,为了克服学习算法性能高度依赖于状态空间探测程度这个问题,本文提出一种动态分层方法,该方法将免疫聚类及二次应答机制融入Sutton提出的Option分层强化学习框架,能对Option状态空间进行动态调整,并沿着学习轨迹动态生成Option内部策略,以二维有障碍栅格空间内两点间最短路径规划为学习任务进行了仿真实验,结果表明,动态分层方法对状态空间探测程度的依赖性很小,动态分层方法更适用于解决大规模强化学习问题.     

基于模糊聚类的分层强化学习算法

本文提出了一种新的分层强化学习Option自动生成算法,以Agent在学习初始阶段探测到的状态空间为输入,采用模糊逻辑神经元的网络进行聚类,在聚类后的各状态子集上通过经验回放学习产生内部策略集,生成Option,仿真实验结果表明了该算法的有效性。     

动态环境中的分层强化学习

现有的强化学习方法都不能很好地处理动态环境中的学习问题,当环境变化时需要重新学习最优策略,若环境变化的时间间隔小于策略收敛时间,学习算法则不能收敛.本文在Option分层强化学习方法的基础上提出一种适应动态环境的分层强化学习方法,该方法利用学习的分层特性,仅关注分层任务子目标状态及当前Option内部环境状态的变化,将策略更新过程限制在规模较小的局部空间或维数较低的高层空间上,从而加快学习速度.以二维动态栅格空间内两点间最短路径规划为背景进行了仿真实验,实验结果表明,该方法策略学习速度明显高于以往的方法,且学习算法收敛性对环境变化频率的依赖性有所降低.     

分层强化学习中的Option自动生成算法

分层强化学习中目前有Option、HAM和MAXQ三种主要方法,其自动分层问题均未得到有效解决,该文针对第一种方法,提出了Option自动生成算法,该算法以Agent在学习初始阶段探测到的状态空间为输入,采用人工免疫网络技术对其进行聚类,在聚类后的各状态子集上通过经验回放学习产生内部策略集,从而生成Option,仿真实验验证了该算法的有效性。     

基于动作空间划分的MAXQ自动分层方法

针对分层强化学习需要人工给出层次结构这一问题,同时考虑到基于状态空间的自动分层方法在环境状态中没有明显子目标时分层效果并不理想的情况,提出一种基于动作空间的自动构造层次结构方法。首先,根据动作影响的状态分量将动作集合划分为多个不相交的子集;然后,分析Agent在不同状态下的可用动作,并识别瓶颈动作;最后,由瓶颈动作与执行次序确定动作子集之间的上下层关系,并构造层次结构。此外,对MAXQ方法中子任务的终止条件进行修改,使所提算法构造的层次结构可以通过MAXQ方法找到最优策略。实验结果表明,所提算法可以自动构造层次结构,而不会受环境变化的干扰。与Q学习、Sarsa算法相比,MAXQ方法根据该结构得到最优策略的时间更短,获得回报更高。验证了所提算法能够有效地自动构造MAXQ层次结构,并使寻找最优策略更加高效。     

基于核密度估计的分层强化学习自动分层算法

分层强化学习中自动分层问题目前仍未得到有效的解决,本文针对Option方法,研究了基于核密度估计的Option自动生成算法,该算法根据分层强化学习的特点并结合改进后的核密度估计层次聚类方法,实现分层强化学习的自动分层,生成子目标,并在此基础上构建出Options。实验结果表明这种算法可以大大加快学习的效率。     

一种新的分层强化学习方法

提出一种集成Option和MAXQ的分层强化学习新方法——OMQ,该方法以MAXQ为基本框架利用先验知识对任务进行人工分层和在线学习,集成Option方法对难以预先细分的子任务进行自动分层。以出租车问题为背景对OMQ学习算法进行了仿真与对比分析,实验结果表明,在任务环境不完全可知条件下,OMQ比Option和MAXQ更适用。     

基于ACCA的Option自动生成算法

提出了一种新的分层强化学习（HRL）Option自动生成算法,以Agent在学习初始阶段探测到的状态空间为输入,并采用改进的蚁群聚类算法（ACCA）对其进行聚类,在聚类后的各状态子集上通过经验回放学习产生内部策略集,从而生成Option,仿真实验验证了该算法是有效的。     

基于概率模型的动态分层强化学习

为解决大规模强化学习中的＂维度灾难＂问题,克服以往学习算法的性能高度依赖于先验知识的局限性,本文提出一种基于概率模型的动态分层强化学习方法.首先基于贝叶斯学习对状态转移概率进行建模,建立基于概率参数的关键状态识别方法,进而通过聚类动态生成若干状态子空间和学习分层结构下的最优策略.仿真结果表明该算法能显著提高复杂环境下智能体的学习效率,适用于未知环境中的大规模学习.     

一种分层强化学习的知识推理方法

针对知识推理过程中,随着推理路径长度的增加,节点的动作空间急剧增长,使得推理难度不断提升的问题,提出一种分层强化学习的知识推理方法(knowledge reasoning method of hierarchical reinforcement learning, MutiAg-HRL),降低推理过程中的动作空间大小。MutiAg-HRL调用高级智能体对知识图谱中的关系进行粗略推理,通过计算下一步关系及给定查询关系之间的相似度,确定目标实体大致位置,依据高级智能体给出的关系,指导低级智能体进行细致推理,选择下一步动作;模型还构造交互奖励机制,对两个智能体的关系和动作选择及时给予奖励,防止模型出现奖励稀疏问题。为验证该方法的有效性,在FB15K-237和NELL-995数据集上进行实验,将实验结果与TransE、MINERVA、HRL等11种主流方法进行对比分析,MutiAg-HRL方法在链接预测任务上的hits@k平均提升了1.85%,MRR平均提升了2%。     

分层强化学习综述

强化学习（reinforcement learning）是机器学习和人工智能领域的重要分支,近年来受到社会各界和企业的广泛关注。强化学习算法要解决的主要问题是,智能体如何直接与环境进行交互来学习策略。但是当状态空间维度增加时,传统的强化学习方法往往面临着维度灾难,难以取得好的学习效果。分层强化学习（hierarchical reinforcement learning）致力于将一个复杂的强化学习问题分解成几个子问题并分别解决,可以取得比直接解决整个问题更好的效果。分层强化学习是解决大规模强化学习问题的潜在途径,然而其受到的关注不高。本文将介绍和回顾分层强化学习的几大类方法。     

Attention control learning in the decision space using state estimation

The main goal of this paper is modelling attention while using it in efficient path planning of mobile robots. The key challenge in concurrently aiming these two goals is how to make an optimal, or near-optimal, decision in spite of time and processing power limitations, which inherently exist in a typical multi-sensor real-world robotic application. To efficiently recognise the environment under these two limitations, attention of an intelligent agent is controlled by employing the reinforcement learning framework. We propose an estimation method using estimated mixture-of-experts task and attention learning in perceptual space. An agent learns how to employ its sensory resources, and when to stop observing, by estimating its perceptual space. In this paper, static estimation of the state space in a learning task problem, which is examined in the Webots^TM simulator, is performed. Simulation results show that a robot learns how to achieve an optimal policy with a controlled cost by estimating the state space instead of continually updating sensory information.     

Hierarchical multi-agent reinforcement learning

In this paper, we investigate the use of hierarchical reinforcement learning (HRL) to speed up the acquisition of cooperative multi-agent tasks. We introduce a hierarchical multi-agent reinforcement learning (RL) framework, and propose a hierarchical multi-agent RL algorithm called Cooperative HRL. In this framework, agents are cooperative and homogeneous (use the same task decomposition). Learning is decentralized, with each agent learning three interrelated skills: how to perform each individual subtask, the order in which to carry them out, and how to coordinate with other agents. We define cooperative subtasks to be those subtasks in which coordination among agents significantly improves the performance of the overall task. Those levels of the hierarchy which include cooperative subtasks are called cooperation levels. A fundamental property of the proposed approach is that it allows agents to learn coordination faster by sharing information at the level of cooperative subtasks, rather than attempting to learn coordination at the level of primitive actions. We study the empirical performance of the Cooperative HRL algorithm using two testbeds: a simulated two-robot trash collection task, and a larger four-agent automated guided vehicle (AGV) scheduling problem. We compare the performance and speed of Cooperative HRL with other learning algorithms, as well as several well-known industrial AGV heuristics. We also address the issue of rational communication behavior among autonomous agents in this paper. The goal is for agents to learn both action and communication policies that together optimize the task given a communication cost. We extend the multi-agent HRL framework to include communication decisions and propose a cooperative multi-agent HRL algorithm called COM-Cooperative HRL. In this algorithm, we add a communication level to the hierarchical decomposition of the problem below each cooperation level. Before an agent makes a decision at a cooperative subtask, it decides if it is worthwhile to perform a communication action. A communication action has a certain cost and provides the agent with the actions selected by the other agents at a cooperation level. We demonstrate the efficiency of the COM-Cooperative HRL algorithm as well as the relation between the communication cost and the learned communication policy using a multi-agent taxi problem.     

Hierarchical reinforcement learning as creative problem solving

Although creativity is studied from philosophy to cognitive robotics, a definition has proven elusive. We argue for emphasizing the creative process (the cognition of the creative agent), rather than the creative product (the artifact or behavior). Owing to developments in experimental psychology, the process approach has become an increasingly attractive way of characterizing creative problem solving. In particular, the phenomenon of insight, in which an individual arrives at a solution through a sudden change in perspective, is a crucial component of the process of creativity.These developments resonate with advances in machine learning, in particular hierarchical and modular approaches, as the field of artificial intelligence aims for general solutions to problems that typically rely on creativity in humans or other animals. We draw a parallel between the properties of insight according to psychology and the properties of Hierarchical Reinforcement Learning (HRL) systems for embodied agents. Using the Creative Systems Framework developed by Wiggins and Ritchie, we analyze both insight and HRL, establishing that they are creative in similar ways. We highlight the key challenges to be met in order to call an artificial system “insightful”.     

基于深度强化学习的三维路径规划算法

合理的路线选择是智能体三维路径规划研究领域的难点。现有路径规划方法存在不能很好地适应未知地形,避障形式单一等问题。针对这些问题,提出了一种基于LSTM-PPO的智能体三维路径规划算法。利用虚拟射线探测仿真环境,并将收集到的状态空间和动作状态引入长短时记忆网络。通过额外的奖惩函数和好奇心驱动让智能体学会跳跃通过低矮障碍物,避开大型障碍物。利用PPO算法的截断项机制使得规划策略更新的幅度更加优化。实验结果表明,该算法是可行的,能够更加智能合理地选择路线,很好地适应存在多样障碍物的未知环境。