ID 原文 译文
56558 近年来, 强化学习方法得到了广泛的关注, 并已在控制领域取得了一定成果. Recently, reinforcement learning (RL) has been studied extensively and achieved promising resultsin a wide range of control tasks.
56559 自主水下机器人(AUV) 能完成复杂多样的水下任务, 是海洋探索开发任务中的关键设备, 强化学习技术的发展为 AUV的智能化控制提供了新的思路.本文着眼于强化学习与 AUV 控制的交叉领域, 重点介绍了强化学习在镇定控制和跟踪控制等底层控制任务方面的应用. In addition, autonomous underwater vehicles (AUVs) are important tools forexecuting complex and challenging tasks in marine environments. The advances in RL offers ample opportunitiesto develop intelligent AUVs.
56560 为此, 我们首先对强化学习框架进行了简要介绍,并在其后对强化学习在 AUV 控制领域面临的挑战以及最新的进展进行了概述.最后, 我们详细介绍了两种针对 AUV 的无模型强化学习控制方法, 希望能够为此类控制器的设计和使用提供参考. This paper provides a selected review of RL-based control for AUVs with a focus onapplications of RL in low-level control tasks for underwater regulation and tracking. We first present a conciseintroduction to the RL-based control framework. We then provide an overview of RL methods for AUV controlproblems, where primary challenges and recent progress are discussed. Finally, two representative RL-basedcontroller cases are presented in detail for model-free RL methods for AUVs.
56561 形式概念分析是以形式背景及其概念格为基础的一种数据分析方法,其中的形式概念明确反映了数据信息中对象与属性间的关系,是哲学中"概念"这一名词的形式化描述. Formal concept analysis, which is based on formal contexts and concept lattices, is a data analysismethod. Formal concepts reflect the relationship between objects and attributes and realize a formal descriptionof concepts in philosophy.
56562 文章提出在形式概念分析框架下进行概念约简的思想,研究保持形式背景中二元关系不变的概念约简的相关理论; To simplify the information description, this paper proposes concept reduction inthe formal concept analysis framework and studies the related theories on concept reduction preserving binaryrelations.
56563 针对概念约简理论中作用不同的3种概念类型,分别从算子角度以及布尔矩阵角度分析了对象(属性)概念的特征,并给出了求解概念约简的方法. In addition, characteristics analysis of three types of concepts that play different roles in reductionprocessing is performed from the perspective of operators and a Boolean matrix. Finally, a method to calculateconcept reduction is provided.
56564 如何提升分布式存储系统中磁盘修复的速度,一直是磁盘修复问题中的难点. Effectively reducing disk repair time for distributed storage systems is an open problem.
56565 优化的途径有两种:一种是通过对解码算法的优化,减少修盘数据的传输量.另外一种方法是通过对修盘过程中数据流的调度,最大化地利用节点的计算能力、传输能力,进而加速修盘进程. Generally,there are two directions, i. e. , constructing better erasure codes to reduce disk repair throughput, and dispatchingrepair tasks to reduce the data flow in switches.
56566 本文从数据流的调度出发,根据数据流图和拓扑结构,计算出了节点的近似最优的修盘数据比例,并依照此比例,设计了分布式存储系统下的近似最优修盘调度算法(NOPT). In this paper, by analyzing the repair data flow, we provide anapproximately optimal disk repair time. We also prove the best ratio of storage nodes for repairing and present theNOPT repair algorithm.
56567 对于主流的两种Reed-Solomon (RS)编码方式,本文做了等价性证明,并给出了编码转换矩阵. We also prove the equivalence of the two decoding algorithms.