检索结果-内蒙古大学图书馆

Sparse Proximal Reinforcement Learning via nested optimization

IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS 2020年第11期50卷 4020-4032页

作者： Song, Tianheng Li, Dazi Jin, Qibing Hirasawa, Kotaro Beijing Univ Chem Technol Inst Automat Beijing 100029 Peoples R China

We consider the tasks of feature selection and policy evaluation based on linear value function approximation in reinforcement learning problems. High-dimension feature vectors and limited number of samples can easily cause over-fitting and computation expensive. To prevent this problem, l(1)-regularized method obtains sparse solutions and thus improves generalization performance. We propose an efficient l(1)-regularized recursive least squares-based online algorithm with O(n(2)) complexity per time-step, termed l(1)-RC. With the help of nested optimization decomposition, l(1)-RC solves a series of standard optimization problems and avoids minimizing mean squares projected Bellman error with l(1)-regularization directly. In l(1)-RC, we propose RC with iterative refinement to minimize the operator error, and we propose an alternating direction method of multipliers with proximal operator to minimize the fixed-point error. The convergence of l(1)-RC is established based on ordinary differential equation method and some extensions are also given. In empirical computations, some state-of-the-art l(1)-regularized methods are chosen as the baselines, and l(1)-RC are tested in both policy evaluation and learning control benchmarks. The empirical results show the effectiveness and advantages ofl(1)-RC.

关键词： Alternating direction method of multipliers (ADMM) l(1)-regularization nested optimization proximal algorithm temporal difference (TD) value function approximation (VFA)

来源：评论

学校读者我要写书评

暂无评论

Online Sparse Temporal Difference Learning Based on nested optimization and Regularized Dual Averaging

引用

IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS 2022年第4期52卷 2042-2052页

作者： Song, Tianheng Li, Dazi Xu, Xin Beijing Univ Chem Technol Dept Automat Beijing 100029 Peoples R China Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha 410073 Peoples R China

In policy evaluation of reinforcement learning tasks, the temporal difference (TD) learning with value function approximation has been widely studied. However, feature representation has a decisive influence on both accuracy of value function approximation and convergence rate. Therefore, it is important to develop the feature selection theory and methods that can efficiently prevent overfitting and improve estimation accuracy in TD learning algorithms. In this article, we propose an online sparse TD learning algorithm for policy evaluation by using l(1)-regualrization for feature selection. The per-step-time runtime computational complexity of the proposed algorithm is linear with respect to feature dimension. The loss function is defined as a nested optimization with l(1)-regularization penalty, and the solver minimizes two suboptimization problems by running stochastic gradient descent and regularized dual averaging method, alternately. The convergence results for the fixed points are also established. The experiments on benchmarks with high-dimensional features show the abilities of learning and generalization of the proposed algorithms.

关键词： l(1)-regularization nested optimization regularized dual averaging (RDA) temporal difference (TD) value function approximation (VFA)

来源：评论

学校读者我要写书评

暂无评论

Parallel asynchronous global search and the nested optimization scheme

引用

JOURNAL OF COMPUTATIONAL ANALYSIS AND APPLICATIONS 2001年第2期3卷 123-145页

作者： Sergeyev, YD Grishagin, VA Univ Calabria ISI CNR DEIS I-87036 Arcavacata Di Rende CS Italy Univ Nizhni Novgorod Software Dept Nizhnii Novgorod 603600 Russia

In this paper, a parallel asynchronous information algorithm for solving multidimensional Lipschitz global optimization problems, where times for evaluating the objective function can be different from point to point, is proposed. This method uses the nested optimization scheme and a new parallel asynchronous global optimization method for solving core univariate subproblems generated by the nested scheme. The properties of the scheme related to parallel computations are investigated. Global convergence conditions for the new method and theoretical conditions of speed up, which can be reached by using asynchronous parallelization in comparison with the pure sequential case, are established. Numerical experiments comparing sequential, synchronous, and asynchronous algorithms are also reported.

关键词： global optimization parallel asynchronous computations nested optimization convergence speed up

来源：评论

学校读者我要写书评

暂无评论

Adaptive granulation Renyi rough entropy image thresholding method with nested optimization

引用

EXPERT SYSTEMS WITH APPLICATIONS 2022年 203卷

作者： Lei, Bo Fan, Jiulun Xian Univ Posts & Telecommun Sch Commun & Informat Engn Xian 710121 Shaanxi Peoples R China Xian Univ Posts & Telecommun Ctr Image & Informat Proc Xian 710121 Shaanxi Peoples R China

The rough entropy based image thresholding algorithm can effectively deal with the uncertainty information in an image. Nevertheless, the existing rough entropy based thresholding methods have two limitations. One is that there is no generalized rough entropy definition. The other is that the granule size cannot be selected automatically in combination with the image information in the granulation process. In order to address these two problems, an adaptive granulation Renyi rough entropy image thresholding method with nested optimization is proposed in this paper. First, a definition of generalized rough entropy with parameter based on Renyi entropy form is proposed to describe the uncertainty information of complex images. Second, in order to select the granule size in combination with the specific image information, a granule size selection method is proposed by maximizing the uniformity of the segmented image regions. Finally, a nested optimization adaptive granulation Renyi rough entropy thresholding segmentation algorithm is proposed. In the experiments, the Renyi rough entropy is compared with the existing four rough entropy in image thresholding. The new parametric rough entropy can obtain better segmentation results than the other existing rough entropy. The novel adaptive granulation Renyi rough entropy thresholding segmentation algorithm is compared with the state-of-the-art image segmentation methods on several different image data sets, which verifies the effectiveness of the novel algorithm.

关键词： Rough set Renyi rough entropy Thresholding segmentation Granulation nested optimization

来源：评论

学校读者我要写书评

暂无评论

optimization of Hybrid Energy Storage Systems for Vehicles with Dynamic On-Off Power Loads Using a nested Formulation

引用

ENERGIES 2018年第10期11卷 2699-2699页

作者： Liu, Jiajun Dong, Huachao Jin, Tianxu Liu, Li Manouchehrinia, Babak Dong, Zuomin Univ Sci & Technol Beijing Sch Mech Engn Beijing 100083 Peoples R China Univ Victoria Dept Mech Engn Victoria BC V8W 2Y2 Canada

In this paper, identification of an appropriate hybrid energy storage system (HESS) architecture, introduction of a comprehensive and accurate HESS model, as well as HESS design optimization using a nested, dual-level optimization formulation and suitable optimization algorithms for both levels of searches have been presented. At the bottom level, design optimization focuses on the minimization of power loss in batteries, converter, and ultracapacitors (UCs), as well as the impact of battery depth of discharge (DOD) to its operation life, using a dynamic programming (DP)-based optimal energy management strategy (EMS). At the top level, HESS optimization of component size and battery DOD is carried out to achieve the minimum life-cycle cost (LCC) of the HESS for given power profiles and performance requirements as an outer loop. The complex and challenging optimization problem is solved using an advanced Multi-Start Space Reduction (MSSR) search method developed for computation-intensive, black-box global optimization problems. An example of load-haul-dump (LHD) vehicles is employed to verify the proposed HESS design optimization method and MSSR leads to superior optimization results and dramatically reduces computation time. This research forms the foundation for the design optimization of HESS, hybridization of vehicles with dynamic on-off power loads, and applications of the advanced global optimization method.

关键词： nested optimization hybrid energy storage system surrogate-based optimization method electrified vehicles

来源：评论

学校读者我要写书评

暂无评论

Learning With nested Scene Modeling and Cooperative Architecture Search for Low-Light Vision

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2023年第5期45卷 5953-5969页

作者： Liu, Risheng Ma, Long Ma, Tengyu Fan, Xin Luo, Zhongxuan Dalian Univ Technol DUT RU Int Sch Informat Sci & Engn Dalian 116024 Liaoning Peoples R China Dalian Univ Technol Sch Software Technol Dalian 116024 Liaoning Peoples R China Peng Cheng Lab Shenzhen 518052 Peoples R China

Images captured from low-light scenes often suffer from severe degradations, including low visibility, color casts, intensive noises, etc. These factors not only degrade image qualities, but also affect the performance of downstream Low-Light Vision (LLV) applications. A variety of deep networks have been proposed to enhance the visual quality of low-light images. However, they mostly rely on significant architecture engineering and often suffer from the high computational burden. More importantly, it still lacks an efficient paradigm to uniformly handle various tasks in the LLV scenarios. To partially address the above issues, we establish Retinex-inspired Unrolling with Architecture Search (RUAS), a general learning framework, that can address low-light enhancement task, and has the flexibility to handle other challenging downstream vision tasks. Specifically, we first establish a nested optimization formulation, together with an unrolling strategy, to explore underlying principles of a series of LLV tasks. Furthermore, we design a differentiable strategy to cooperatively search specific scene and task architectures for RUAS. Last but not least, we demonstrate how to apply RUAS for both low- and high-level LLV applications (e.g., enhancement, detection and segmentation). Extensive experiments verify the flexibility, effectiveness, and efficiency of RUAS.

关键词： Task analysis Computer architecture Training Visualization Deep learning Noise reduction Image enhancement Low-light vision nested optimization Retinex-inspired unrolling cooperative architecture search

来源：评论

学校读者我要写书评

暂无评论

A classification of methods for distributed system optimization based on formulation structure

引用

STRUCTURAL AND MULTIDISCIPLINARY optimization 2009年第5期39卷 503-517页

作者： Tosserams, S. Etman, L. F. P. Rooda, J. E. Eindhoven Univ Technol Dept Mech Engn NL-5600 MB Eindhoven Netherlands

This paper presents a classification of formulations for distributed system optimization based on formulation structure. Two main classes are identified: nested formulations and alternating formulations. nested formulations are bilevel programming problems where optimization subproblems are nested in the functions of a coordinating master problem. Alternating formulations iterate between solving a master problem and disciplinary subproblems in a sequential scheme. Methods included in the former class are collaborative optimization and BLISS2000. The latter class includes concurrent subspace optimization, analytical target cascading, and augmented Lagrangian coordination. Although the distinction between nested and alternating formulations has not been made in earlier comparisons, it plays a crucial role in the theoretical and computational properties of distributed optimization methods. The most prominent general characteristics for each class are discussed in more detail, providing valuable insights for the theoretical analysis and further development of distributed optimization methods.

关键词： Distributed optimization Classification nested optimization Alternating optimization Multidisciplinary design optimization Multi-level optimization Bilevel programming

来源：评论

学校读者我要写书评

暂无评论

A new deregulated demand response scheme for load over-shifting city in regulated power market

引用

APPLIED ENERGY 2022年 311卷 118337-118337页

作者： Xu, Fangyuan Zhu, Weidong Wang, Yu Fei Lai, Chun Sing Yuan, Haoliang Zhao, Yujia Guo, Siming Fu, Zhengxin Guangdong Univ Technol Sch Automat Dept Elect Engn Guangzhou 510000 Peoples R China Southeast Univ Sch Elect Engn Nanjing 210096 Peoples R China Brunel Univ London Dept Elect & Elect Engn Brunel Interdisciplinary Power Syst Res Ctr London UB8 3PH England

Time-based demand response (DR) is an effective way to improve the reliability of power grid and reduce energy costs. Time-Of-Use tariff (TOU) has been adopted by many countries and achieved good performance. However, in cities with a large proportion of industrial consumers, load over-shifting phenomenon leads to new peak electricity consumption and reduces the effect of TOU. This paper proposes a new deregulated demand response scheme (DDR) to solve the load over-shifting problem. The scheme selects industrial consumers with large shiftable load in the city as load adjustment component and provides independent tariff to each consumer. Different to other methods with the requirement of entire scheme replacement, the proposed DDR only influences a small group of consumers with much lower implementation risk. Also, the cost of consumers and profit of agent can be improved at the same time as shown in the numerical study. In the proposed scheme, the interests of consumers and the agent need to be considered in the formulation of independent tariff, which forms a nested optimization problem that is difficult to solve quickly. In this paper, a novel and efficient approximate algorithm is proposed to solve the optimization problem. The proposed algorithm can produce optimal solutions similar to Genetic Algorithm with higher computational efficiency.

关键词： Load over-shifting Deregulated demand response Independent tariff nested optimization

来源：评论

学校读者我要写书评

暂无评论

Wake losses optimization of offshore wind farms with moveable floating wind turbines

引用

ENERGY CONVERSION AND MANAGEMENT 2015年 89卷 933-941页

作者： Rodrigues, S. F. Pinto, R. Teixeira Soleimanzadeh, M. Bosman, Peter A. N. Bauer, P. Delft Univ Technol DC Syst & Storage Grp NL-2628 CD Delft Netherlands ECN NL-1755 LE Petten Netherlands CWI NL-1098 XG Amsterdam Netherlands

In the future, floating wind turbines could be used to harvest energy in deep offshore areas where higher wind mean speeds are observed. Currently, several floating turbine concepts are being designed and tested in small scale projects;in particular, one concept allows the turbine to move after installation. This article presents a novel layout optimization framework for wind farms composed of moveable floating turbines. The proposed framework uses an evolutionary optimization strategy in a nested configuration which simultaneously optimizes the anchoring locations and the wind turbine position within the mooring lines for each individual wind direction. The results show that maximum energy production is obtained when moveable wind turbines are deployed in an optimized layout. In conclusion, the framework represents a new design optimization tool for future offshore wind farms composed of moveable floating turbines. (C) 2014 Elsevier Ltd. All rights reserved.

关键词： Moveable floating turbines nested optimization Offshore wind energy Turbine micro siting Wake losses

来源：评论

学校读者我要写书评

暂无评论

Feasibility Study of a Rapid Evaluate and Adjust Device (READ) for Custom Foot Orthoses Prescription

引用

IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING 2020年第8期28卷 1760-1770页

作者： Qaiser, Zeeshan Faraz, Ahmad Johnson, Shane Shanghai Jiao Tong Univ Univ Michigan Shanghai Jiao Tong Univ Joint Inst Adv Intelligent Mat & Struct AIMS Minhang Campus Shanghai 200240 Peoples R China Shanghai Jiao Tong Univ State Key Lab Mech Syst & Vibrat Shanghai 200240 Peoples R China

Objective: Custom foot orthoses (CFOs) are typically used for the prevention and cure of lower extremity injuries (LEIs). Typically, CFOs are designed and prescribed based on iterative loops including: (1) follow-up loops between the patient and the physician, and (2) design feedback loops between the physician and the fabricator. The current prescription methodology has some deficiencies, i.e. excessive time to satisfactory treatment, and low repeatability in custom fabrication because of missing alignment, soft tissue considerations, and subjective feedback. There are significant opportunities to develop a new CFOs prescription procedure which can improve accuracy prior to the fabrication process by reducing time through minimizing iterations. Methods: First, a novel "rapid evaluate and adjust device" (READ) prescription methodology is proposed for CFO design by combining the follow-up loops with design feedback loop. To support the idea of the READ prescription method a novel 3D ergonomic measurement system (3DEMS) is developed. The 3DEMS is designed for the following key targets to: (1) improve the communication between the patient/physician and the fabricator, (2) reduce time to satisfactory treatment, (3) improve repeatability by considering the alignment and the soft tissue deformations, (4) archive digitally with minimal data, (5) reduce the system complexity, and (6) validate with plantar pressure measurements (i.e. Novel Pedar(R)). The design process of the 3DEMS involved the following steps: (1) 3D data collection at the desired loading, (2) nested optimization to determine optimal segment design, and (3) system fabrication considering alignment and feedback control. Results: The results show that the READ prescription method with 3DEMS can be used to recreate the medial longitudinal arch for a range of arch height indices (AHI) by using a minimal number of parameters i.e. 6 parameters, and significant increases in mean peak pressure are observed between

关键词： Custom foot orthoses plantar shape measurement nested optimization

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：