检索结果-南通市图书馆

Efficient reinforcement learning in continuous state and action spaces with Dyna and policy approximation

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2019年第1期13卷 106-126页

作者： Shan ZHONG Quan LIU Zongzhang ZHANG qiming fu School of Computer Science and Technology Soochow UniversitySuzhou 215000China School of Computer Science and Engineering Changshu Institute of TechnologyChangshu 215500China Jiangsu Province Key Laboratory of Intelligent Building Energy Efficiency Suzhou University of Science and TechnologySuzhou 215006China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun 130012China Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing 210000China College of Electronic &Information Engineering Suzhou University of Science and TechnologySuzhou 215000China

Dyna is an effective reinforcement learning (RL) approach that combines value function evaluation with model ***,existing works on Dyna mostly discuss only its efficiency in RL problems with discrete action *** paper proposes a novel Dyna variant,called Dyna-LSTD-PA,aiming to handle problems with continuous action ***-LSTD-PA stands for Dyna based on least-squares temporal difference (LSTD)and policy ***-LSTD-PA consists of two simultaneous,interacting *** learning process determines the probability distribution over action spaces using the Gaussian distribution;estimates the underlying value function,policy,and model by linear representation;and updates their parameter vectors online by LSTD(,t).The planning process updates the parameter vector of the value function again by using ofttine LSTD(2).Dyna-LSTD-PA also uses the Sherman-Morrison formula to improve the efficiency of LSTD(,t),and weights the parameter vector of the value function to bring the two processes ***,the global error bound is derived by considering approximation,estimation,and model ***,Dyna-LSTD-PA outperforms two representative methods in terms of convergence rate,success rate,and stability performance on four benchmark RL problems.

关键词： problem solving control methods heuristic search methods dynamic programming

Experience Replay for Least-Squares Policy Iteration

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2014年第3期1卷 274-281页

作者： Quan Liu Xin Zhou Fei Zhu qiming fu Yuchen fu the School of Computer Science and Technology Soochow University the Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University the School of Computer Science and Technology Soochow University

Policy iteration,which evaluates and improves the control policy iteratively,is a reinforcement learning *** evaluation with the least-squares method can draw more useful information from the empirical data and therefore improve the data ***,most existing online least-squares policy iteration methods only use each sample just once,resulting in the low utilization *** the goal of improving the utilization efficiency,we propose an experience replay for least-squares policy iteration(ERLSPI)and prove its *** method combines online least-squares policy iteration method with experience replay,stores the samples which are generated online,and reuses these samples with least-squares method to update the control *** apply the ERLSPI method for the inverted pendulum system,a typical benchmark *** experimental results show that the method can effectively take advantage of the previous experience and knowledge,improve the empirical utilization efficiency,and accelerate the convergence speed.

关键词： Learning (artificial intelligence) Markov processes Computational efficiency Convergence Least squares methods

Multi-label Image Classification via Coarse-to-Fine Attention

在线全文

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2019年第6期28卷 1118-1126页

作者： LYU Fan LI Linyan Victor S.Sheng fu qiming HU fuyuan School of Electronic and Information Engineering Suzhou University of Science and Technology College of Intelligence and Computing Tianjin University Suzhou Institute of Trade and Commerce Department of Computer Science Texas Tech University

Great efforts have been made by using deep neural networks to recognize multi-label *** multi-label image classification is very complicated,many studies seek to use the attention mechanism as a kind of guidance. Conventional attention-based methods always analyzed images directly and aggressively, which is difficult to well understand complicated scenes. We propose a global/local attention method that can recognize a multi-label image from coarse to fine by mimicking how human-beings observe images. Our global/local attention method first concentrates on the whole image, and then focuses on its local specific objects. We also propose a joint max-margin objective function, which enforces that the minimum score of positive labels should be larger than the maximum score of negative labels horizontally and vertically. This function further improve our multi-label image classification method. We evaluate the effectiveness of our method on two popular multi-label image datasets(i.e., Pascal VOC and MS-COCO). Our experimental results show that our method outperforms state-of-the-art methods.

关键词： Multi-label classification Convolutional neural network Recurrent neural network Attention

MAML^(2):meta reinforcement learning via meta-learning for task categories

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2023年第4期17卷 57-67页

作者： qiming fu Zhechao WANG Nengwei FANG Bin XING Xiao ZHANG Jianping CHEN School of Electronics and Information Engineering Suzhou University of Science and TechnologySuzhou 215009China Chongqing Industrial Big Data Innovation Center Co.Ltd. Chongqing 404100China School of Medical Informatics Xuzhou Medical UniversityXuzhou 221004China School of Architecture and Urban Planning Suzhou University of Science and TechnologySuzhou 215009China

Meta-learning has been widely applied to solving few-shot reinforcement learning problems,where we hope to obtain an agent that can learn quickly in a new ***,these algorithms often ignore some isolated tasks in pursuit of the average performance,which may result in negative adaptation in these isolated tasks,and they usually need sufficient learning in a stationary task *** this paper,our algorithm presents a hierarchical framework of double meta-learning,and the whole framework includes classification,meta-learning,and ***,in the classification process,we classify tasks into several task subsets,considered as some categories of tasks,by learned parameters of each task,which can separate out some isolated tasks ***,in the meta-learning process,we learn category parameters in all subsets via ***,based on the gradient of each category parameter in each subset,we use meta-learning again to learn a new metaparameter related to the whole task set,which can be used as an initial parameter for the new ***,in the re-adaption process,we adapt the parameter of the new task with two steps,by the meta-parameter and the appropriate category parameter ***,we demonstrate our algorithm prevents the agent from negative adaptation without losing the average performance for the whole task ***,our algorithm presents a more rapid adaptation process within ***,we show the good performance of our algorithm with fewer samples as the agent is exposed to an online meta-learning setting.

关键词： meta-learning reinforcement learning few-shot learning negative adaptation

HVAC Optimal Control Based on the Sensitivity Analysis:An Improved SA Combination Method Based on a Neural Network

在线全文

学校读者我要写书评

暂无评论

Computer Modeling in Engineering & Sciences 2023年第9期136卷 2741-2758页

作者： Lifan Zhao Zetian Huang qiming fu Nengwei Fang Bin Xing Jianping Chen School of Electronic and Information Engineering Suzhou University of Science and TechnologySuzhou215009China Jiangsu Key Laboratory of Intelligent Building Energy Efficiency Suzhou University of Science and TechnologySuzhou215009China Suzhou Key Laboratory of Mobile Network Technology and Application Suzhou University of Science and TechnologySuzhou215009China Chongqing Industrial Big Data Innovation Center Co. Ltd.Chongqing400707China

Aiming at optimizing the energy consumption of HVAC,an energy conservation optimization method was proposed for HVAC systems based on the sensitivity analysis(SA),named the sensitivity analysis combination method(SAC).Based on the SA,neural network and the related settings about energy conservation of HVAC systems,such as cooling water temperature,chilled water temperature and supply air temperature,were ***,based on the data of the existing HVAC system,various optimal control methods ofHVAC systems were tested and evaluated by a simulated HVAC system in *** results show that the proposed SA combination method can reduce significant computational load while maintaining an equivalent energy performance compared with traditional optimal control methods.

关键词： Energy conservation sensitivity analysis HVAC system neural network

MAQMC:Multi-Agent Deep Q-Network for Multi-Zone Residential HVAC Control

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computer Modeling in Engineering & Sciences 2023年第9期136卷 2759-2785页

作者： Zhengkai Ding qiming fu Jianping Chen You Lu Hongjie Wu Nengwei Fang Bin Xing School of Electronic and Information Engineering Suzhou University of Science and TechnologySuzhou215009China Jiangsu Province Key Laboratory of Intelligent Building Energy Efficiency Suzhou University of Science and TechnologySuzhou215009China School of Architecture and Urban Planning Suzhou University of Science and TechnologySuzhou215009China Chongqing Industrial Big Data Innovation Center Co. Ltd.Chongqing400707China

The optimization of multi-zone residential heating,ventilation,and air conditioning(HVAC)control is not an easy task due to its complex dynamic thermal model and the uncertainty of occupant-driven cooling *** reinforcement learning(DRL)methods have recently been proposed to address the HVAC control ***,the application of single-agent DRL formulti-zone residential HVAC controlmay lead to non-convergence or slow *** this paper,we propose MAQMC(Multi-Agent deep Q-network for multi-zone residential HVAC Control)to address this challenge with the goal of minimizing energy consumption while maintaining occupants’thermal *** is divided into MAQMC2(MAQMC with two agents:one agent controls the temperature of each zone,and the other agent controls the humidity of each zone)and MAQMC3(MAQMC with three agents:three agents control the temperature and humidity of three zones,respectively).The experimental results showthatMAQMC3 can reduce energy consumption by 6.27%andMAQMC2 by 3.73%compared with the fixed point;compared with the rule-based,MAQMC3 andMAQMC2 respectively can reduce 61.89%and 59.07%comfort *** addition,experiments with different regional weather data demonstrate that the well-trained MAQMC RL agents have the robustness and adaptability to unknown environments.

关键词： Deep reinforcement learning multi-zone residential HVAC multi-agent energy conservation comfort

一种用于连续动作空间的最小二乘行动者-评论家方法

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

计算机研究与发展 2014年第3期51卷 548-558页

作者：朱斐刘全傅启明伏玉琛苏州大学计算机科学与技术学院江苏苏州215006 苏州大学系统生物学研究中心江苏苏州215006 符号计算与知识工程教育部重点实验室(吉林大学) 长春130012

解决具有连续动作空间的问题是当前强化学习领域的一个研究热点和难点.在处理这类问题时,传统的强化学习算法通常利用先验信息对连续动作空间进行离散化处理,然后再求解最优策略.然而,在很多实际应用中,由于缺乏用于离散化处理的先验信... 详细信息

解决具有连续动作空间的问题是当前强化学习领域的一个研究热点和难点.在处理这类问题时,传统的强化学习算法通常利用先验信息对连续动作空间进行离散化处理,然后再求解最优策略.然而,在很多实际应用中,由于缺乏用于离散化处理的先验信息,算法效果会变差甚至算法失效.针对这类问题,提出了一种最小二乘行动者-评论家方法(least square actor-critic algorithm,LSAC),使用函数逼近器近似表示值函数及策略,利用最小二乘法在线动态求解近似值函数参数及近似策略参数,以近似值函数作为评论家指导近似策略参数的求解.将LSAC算法用于解决经典的具有连续动作空间的小车平衡杆问题和mountain car问题,并与Cacla(continuous actor-critic learning automaton)算法和eNAC(episodic natural actor-critic)算法进行比较.结果表明,LSAC算法能有效地解决连续动作空间问题,并具有较优的执行性能.

关键词：强化学习行动者-评论家算法连续动作空间最小二乘法小车平衡杆问题 mountain car问题

东北国有林区剩余劳动力转移就业的障碍与突破

在线全文

学校读者我要写书评

暂无评论

林业经济问题 2021年第6期41卷 561-568页

作者：林红付启明闽江学院马克思主义学院福州350108 黑龙江八一农垦大学经济管理学院黑龙江大庆163319

利用2014-2016年东北国有林区跟踪调查数据分析东北国有林区剩余劳动力转移就业的障碍因素。结果表明:东北国有林区剩余劳动力转移就业主要受到劳动力因素、产业因素和制度因素的制约,解决剩余劳动力转移就业问题,需要在提升劳动力就业... 详细信息

利用2014-2016年东北国有林区跟踪调查数据分析东北国有林区剩余劳动力转移就业的障碍因素。结果表明:东北国有林区剩余劳动力转移就业主要受到劳动力因素、产业因素和制度因素的制约,解决剩余劳动力转移就业问题,需要在提升劳动力就业能力的同时,还需要政府的政策扶持与组织保障。

关键词：东北国有林区剩余劳动力转移就业

在线全文

学校读者我要写书评

暂无评论

基于函数逼近的冗余值迭代算法

模式识别与人工智能 2017年第7期30卷 663-672页

作者：陈建平胡文傅启明苏州科技大学电子与信息工程学院苏州215009 苏州科技大学江苏省建筑智慧节能重点实验室苏州215009 苏州科技大学苏州市移动网络技术与应用重点实验室苏州215009 吉林大学符号计算与知识工程教育部重点实验室长春130012

针对值迭代算法存在算法收敛不稳定及收敛速度慢的问题,文中提出改进的基于函数逼近的冗余值迭代算法.结合值迭代算法与贝尔曼冗余值迭代算法,引入权重因子,构建值函数参数更新向量.同时从理论上证明,利用此更新向量更新值函数参数可以... 详细信息

针对值迭代算法存在算法收敛不稳定及收敛速度慢的问题,文中提出改进的基于函数逼近的冗余值迭代算法.结合值迭代算法与贝尔曼冗余值迭代算法,引入权重因子,构建值函数参数更新向量.同时从理论上证明,利用此更新向量更新值函数参数可以保证算法收敛,解决值迭代算法收敛不稳定的问题.此外,算法引入遗忘因子,加快权重向量的更新速率和算法收敛速度.在Grid World问题上的实验表明,文中算法收敛性能较好,具有较好的鲁棒性.

关键词：强化学习值迭代函数逼近梯度下降贝尔曼冗余