基于参数探索的期望最大化策略搜索

针对随机探索易于导致梯度估计方差过大的问题,提出一种基于参数探索的期望最大化(Expectation-maximization,EM)策略搜索方法.首先,将策略定义为控制器参数的一个概率分布.然后,根据定义的概率分布直接在控制器参数空间进行多次采样以收集样本.在每一幕样本的收集过程中,由于选择的动作均是确定的,因此可以减小采样带来的方差,从而减小梯度估计方差.最后,基于收集到的样本,通过最大化期望回报函数的下界来迭代地更新策略参数.为减少采样耗时和降低采样成本,此处利用重要采样技术以重复使用策略更新过程中收集的样本.两个连续空间控制问题的仿真结果表明,与基于动作随机探索的策略搜索强化学习方法相比,本文所提方法不仅学到的策略最优,而且加快了算法收敛速度,具有较好的学习性能.     

基于随机方差减小方法的DDPG算法

针对深度确定性策略梯度算法（DDPG）收敛速度比较慢,训练不稳定,方差过大,样本应用效率低的问题,提出了一种基于随机方差减小梯度方法的深度确定性策略梯度算法（SVR-DDPG）。该算法通过利用随机方差减小梯度技术（SVRG）提出一种新的创新优化策略,将之运用到DDPG算法之中,在DDPG算法的参数更新过程中,加入了随机方差减小梯度技术,利用该方法的更新方式,使得估计的梯度方差有一个不断减小的上界,令方差不断缩小,从而在小的随机训练子集的基础上找到更加精确的梯度方向,以此来解决了由近似梯度估计误差引发的问题,加快了算法的收敛速度。将SVR-DDPG算法以及DDPG算法应用于Pendulum和Mountain Car问题,实验结果表明,SVR-DDPG算法具有比原算法更快的收敛速度,更好的稳定性,以此证明了算法的有效性。     

一类非线性动态系统基于强化学习的最优控制制

提出一类非线性不确定动态系统基于强化学习的最优控制方法. 该方法利用欧拉强化学习算法估计对象的未知非线性函数, 给出了强化学习中回报函数和策略函数迭代的在线学习规则. 通过采用向前欧拉差分迭代公式对学习过程中的时序误差进行离散化, 实现了对值函数的估计和控制策略的改进. 基于值函数的梯度值和时序误差指标值, 给出了该算法的步骤和误差估计定理. 小车爬山问题的仿真结果表明了所提出方法的有效性.

     

基于TD(λ)的自然梯度强化学习算法

近年来强化学习中的策略梯度方法以其良好的收敛性能吸引了广泛的关注。研究了平均模型中的自然梯度算法,针对现有算法估计梯度时效率较低的问题,在梯度估计的值函数逼近中采用了TD(λ)方法。TD(λ)中的资格迹使学习经验的传播更加高效,从而能够降低梯度估计的方差,提升算法的收敛速度。车杆平衡系统仿真实验验证了所提算法的有效性。     

AdaSVRG：自适应学习率加速SVRG

在深度学习任务中,随机方差衰减梯度法通过降低随机梯度方差,因此,其具有较好的稳定性和较高的计算效率。然而,这类方法在学习过程中均使用恒定的学习率,降低了随机方差衰减梯度法的计算效率。基于随机方差衰减梯度法,借鉴动量加速思想并对梯度估计采取加权平均策略,对学习率利用历史梯度信息进行自动调整,提出了自适应随机方差衰减梯度法。基于MNIST和CIFAR-10数据集,验证提出的自适应随机方差衰减梯度法的有效性。实验结果表明,自适应随机方差衰减梯度法在收敛速度和稳定性方面优于随机方差衰减梯度法和随机梯度下降法。     

一种解决连续空间问题的真实在线自然梯度AC算法

策略梯度作为一种能有效解决连续空间决策问题的方法被广泛研究.然而,由于在策略估计过程中存在较大的方差,因此基于策略梯度的方法往往受到样本利用率低、收敛速度慢等限制.针对该问题,提出了真实在线增量式自然梯度行动者-评论家算法（TOINAC）.TOINAC算法采用优于传统梯度的自然梯度,在真实在线时间差分（TOTD）算法的基础上,提出了一种新型的向前观点,改进了自然梯度行动者-评论家算法.在评论家部分,利用TOTD算法高效性的特点来估计值函数;在行动者部分,引入一种新的向前观点来估计自然梯度,再利用资格迹将自然梯度估计变为在线估计,提高了自然梯度估计的准确性和算法的效率.将TOINAC算法与核方法以及正态策略分布结合,解决连续空间问题.最后,在平衡杆、Mountain Car、以及Acrobot等连续问题上进行了仿真实验,验证算法的有效性.     

融合序列模式评分的策略梯度推荐算法

推荐算法在一定程度上解决了信息过载问题,但传统推荐模型在挖掘数据特性方面有待改进。为此,结合强化学习方法提出一种融合序列模式评分的策略梯度推荐算法。将推荐过程建模为马尔可夫决策过程;分析推荐基础数据特性模式,设计以序列模式评分为奖励的反馈函数,在算法的每一次迭代过程中学习;通过对累积奖励设计标准化操作来降低策略梯度的方差。将该方法应用到电影推荐中进行验证,结果表明所提方法具有较好的推荐准确性。     

求解非光滑强凸优化问题的减小方差加权随机算法*

在光滑问题随机方法中使用减小方差策略,能够有效改善算法的收敛效果.文中同时引用加权平均和减小方差的思想,求解“L1+L2+Hinge”非光滑强凸优化问题,得到减小方差加权随机算法(α-HRMDVR-W).在每步迭代过程中使用减小方差策略,并且以加权平均的方式输出,证明其具有最优收敛速率,并且该收敛速率不依赖样本数目.与已有减小方差方法相比,α-HRMDVR-W每次迭代中只使用部分样本代替全部样本修正梯度.实验表明α-HRMDVR-W在减小方差的同时也节省CPU时间.     

不动点视角下的强化学习算法综述

近年来,强化学习已成为求解序贯决策任务的范式.然而,在实际应用中,强化学习算法仍存在三个问题:(1)什么解最优?(2)如何保证算法的稳定性?(3)如何加速算法的收敛?本文从不动点视角总结了强化学习算法的设计原理.首先,分析了值函数估计最优解与可行解的构造问题;其次,根据Banach不动点定理和Lyapunov第二判定定理,总结了已有基于值函数强化学习算法的稳定性问题,包括基于表格、线性估计、非线性估计、非参估计等值函数的算法在同策略和异策略情况下的收敛性;然后,从不动点的偏差与方差控制角度,解读了多种提高算法准确性或收敛速度的改进思想;最后总结和展望了强化学习算法的改进方向.     

基于元学习的自适应视频流算法

针对现有的码率自适应(adaptive bitrate, ABR)算法存在控制规则简单，不能有效提升用户体验质量(quality of experience, QoE),提出一种基于元学习的LABR(reinforcement learning based ABR)算法。采用策略梯度训练策略网络，利用元学习(meta-learning)方法学习基线(baseline)函数来减少因网络吞吐量差异产生的方差，进一步提高模型的准确性和鲁棒性；通过在策略函数中加入熵损失方法提高累计期望奖励值。实验结果表明，LABR算法具有泛化性与鲁棒性，能有效提高用户的视频体验质量。     

Call for Papers The 6^th International Conference on Grid and Cooperative Computing

Recent years have seen rapid advances in various grid-related technologies, middleware, and applications. The GCC conference has become one of the largest scientific events worldwide in grid and cooperative computing. The 6th international conference on grid and cooperative computing (GCC2007) Sponsored by China Computer Federation (CCF),Institute of Computing Technology, Chinese Academy of Sciences (ICT) and Xinjiang University ,and in Cooperation with IEEE Computer Soceity ,is to be held from August 16 to 18, 2007 in Urumchi, Xinjiang, China.     

煤矿井下安全监控分站的设计及其仿真实现

为了设计一种具有低成本、低功耗、易操作、功能强且可靠性高的煤矿井下安全分站,针对煤矿安全生产实际,文章提出了采用MCS-51系列单片机为核心、具有CAN总线通信接口的煤矿井下安全监控分站的设计方案;首先给出煤矿井下安全监控分站的整体构架设计,然后着重阐述模拟量输入信号处理系统的设计过程,最后说明单片机最小系统及其键盘、显示、报警、通信等各个组成部分的设计;为验证设计方案的可行性与有效性,使用Proteus软件对设计内容进行仿真验证,设计的煤矿井下安全监控分站具有瓦斯、温度等模拟量参数超标报警功能和电机开停、风门开闭等开关量指示功能;仿真结果表明:设计的煤矿井下安全监控分站具有一定的实际应用价值.     

法律信息库系统

本文分析了法律数据库的结构和特点，介绍了采用面向对象设计方法和超文本数据库技术开发和实现法律信息库系统将作为重要网络资源之一为不同用户进行法律咨询服务。     

View-based model-driven architecture for enhancing maintainability of data access services

In modern service-oriented architectures, database access is done by a special type of services, the so-called data access services (DAS). Though, particularly in data-intensive applications, using and developing DAS are very common today, the link between the DAS and their implementation, e.g. a layer of data access objects (DAOs) encapsulating the database queries, still is not sufficiently elaborated, yet. As a result, as the number of DAS grows, finding the desired DAS for reuse and/or associated documentation can become an impossible task. In this paper we focus on bridging this gap between the DAS and their implementation by presenting a view-based, model-driven data access architecture (VMDA) managing models of the DAS, DAOs and database queries in a queryable manner. Our models support tailored views of different stakeholders and are scalable with all types of DAS implementations. In this paper we show that our view-based and model driven architecture approach can enhance software development productivity and maintainability by improving DAS documentation. Moreover, our VMDA opens a wide range of applications such as evaluating DAS usage for DAS performance optimization. Furthermore, we provide tool support and illustrate the applicability of our VMDA in a large-scale case study. Finally, we quantitatively prove that our approach performs with acceptable response times.     

International Journal of Automation and Computing Table of Contents

正~~     

Information for authors

正SCIENCE CHINA Information Sciences(Sci China Inf Sci),cosponsored by the Chinese Academy of Sciences and the National Natural Science Foundation of China,and published by Science China Press,is committed to publishing highquality,original results of both basic and applied research in all areas of information sciences,including computer science and technology;systems science,control science and engineering(published in Issues with odd numbers);information and communication engineering;electronic science and technology(published in Issues with even numbers).Sci China Inf Sci is published monthly in both print and electronic forms.It is indexed by Academic OneFile,Astrophysics Data System(ADS),CSA,Cabells,Current Contents/Engineering,Computing and Technology,DBLP,Digital Mathematics Registry,Earthquake Engineering Abstracts,Engineering Index,Engineered Materials Abstracts,Gale,Google,INSPEC,Journal Citation Reports/Science Edition,Mathematical Reviews,OCLC,ProQuest,SCOPUS,Science Citation Index Expanded,Summon by Serial Solutions,VINITI,Zentralblatt MATH.     

Erratum to: Fast global kernel fuzzy c-means clustering algorithm for consonant/vowel segmentation of speech signal

正Erratum to:J Zhejiang Univ-Sci C(ComputElectron)2014 15(7):551-563doi:10.1631/jzus.C1300320The original version of this article unfortunately contained mistakes.Algorithm 6 should be as follows:Algorithm 6 FGKFCM-F clustering Input:(1)X={x_1,x_2,…,x_N},,x_iR~d,i=1,2,…,N,the dataset;(2)C,1C≤N,the number of clusters;(3)ε0,the stopping criterion;     

Journal of Zhejiang University-SCIENCE C (Computers & Electronics) Total contents

正~~     

A pragmatic multi-method investigation of discrepant technological events: Coping,attributions, and ‘accidental’ learning

Discrepant technological events or situations that entail a problem, a misunderstanding or a difficulty with the Information Technology (IT) being employed, are common in the workplace, and can lead to frustration and avoidance behaviors. Little is known, however, about how individuals cope with these events. This paper examines these events by using a multi-method pragmatic approach informed by coping theory. The results of two studies – a critical incident study and an experiment – serve to build and test, respectively, a theoretical model that posits that individuals use a variety of strategies when dealing with these events: they experience negative emotions, make external attributions, and adopt engagement coping strategies directed at solving the event, eventually switching to a disengagement coping strategy when they feel they have no control over the situation. Furthermore, users’ efforts may result in ‘accidental’ learning as they try to overcome the discrepant IT events through engagement coping. The paper ends with a discussion of the results in light of existing literature, future opportunities for research, and implications for practice.