检索结果-内蒙古大学图书馆

An Ensemble Surrogate-Based Coevolutionary Algorithm for Solving Large-Scale Expensive Optimization Problems

IEEE TRANSACTIONS ON CYBERNETICS 2023年第9期53卷 5854-5866页

作者： Wu, Xunfeng Lin, Qiuzhen Li, Jianqiang Tan, Kay Chen Leung, Victor C. M. Shenzhen Univ Coll Comp Sci & Software Engn Shenzhen 518060 Peoples R China Hong Kong Polytech Univ Dept Comp Hong Kong Peoples R China

Surrogate-assisted evolutionary algorithms (SAEAs) have shown promising performance for solving expensive optimization problems (EOPs) whose true evaluations are computationally or physically expensive. However, most existing SAEAs only focus on the problems with low dimensionality and they rarely consider solving large-scale EOPs (LSEOPs). To fill this research gap, this article proposes an ensemble surrogate-based coevolutionary optimizer for tackling LSEOPs. First, some local surrogate models are trained with low-dimensional data subsets by using feature selection on the large-scale decision variables, a part of which are used to build a selective ensemble surrogate for better approximating the target LSEOP. Then, a coevolutionary optimizer guided by the ensemble surrogate is designed by running two populations to cooperatively solve the target LSEOP and the simplified auxiliary problem. The information of offspring from the two populations is shared to facilitate the coevolution process, which can exploit the searching experience from the simplified auxiliary problem to help solving the target LSEOP. Finally, an effective infill selection criterion is used to update the ensemble surrogate and enhance its approximate performance. To evaluate the performance of the proposed algorithm, a number of well-known benchmark problems are used and the experimental results validate our superior performance over nine state-of-the-art SAEAs on most cases.

关键词： Predictive models Optimization approximation algorithms Search problems Data models Computational modeling Support vector machines Ensemble surrogate large-scale expensive optimization problem (LSEOP) surrogate-assisted evolutionary algorithm (SAEA)

来源：评论

学校读者我要写书评

暂无评论

An Optimal Nonlinear Type-2 Fuzzy FOPID Control Design Based on Integral Performance Criteria Using FSM

引用

IEEE ACCESS 2023年 11卷 53439-53467页

作者： Mohammad, M. Al-Momani Al-Mbaideen, Amneh Al-Odienat, Abdullah I. Alawasa, Khaled Mohammad Al-Gharaibeh, Saba F. Mutah Univ Dept Elect Engn Mutah 61710 Jordan Sultan Qaboos Univ Dept Elect & Comp Engn Muscat 123 Oman

A fractional-order fuzzy proportional integral derivative (PID) controller is a controller that combines the benefits of fractional calculus and fuzzy logic with the conventional PID controller. In this paper, a four-stage optimization algorithm is proposed for the design of a Type-2 Fuzzy fractional-order PID controller based on the Fourier Series Method (FSM). Three distinct control structures are introduced: Type-2 fuzzy fractional PD + fractional PI controller, Type-2 fuzzy fractional PID, and Type-2 fuzzy fractional PD + Type-2 fuzzy fractional PI controller. In addition to a modified multi-performance criterion cost function, four integral performance criteria are employed as cost functions for each stage. The suggested algorithm avoids the utilization of the approximation equivalent for the fractional-order system and instead employs FSM. Furthermore, the approach optimizes the nonlinearity within the upper membership function (UMF) and the uncertainty range through the lower membership function, as opposed to arbitrary selection. By considering variations in the membership functions, the outcomes exhibit a superior response compared to previous investigations. The results of the three control structures are compared with the traditional PID controller, and simulation results demonstrate the feasibility of this technique. The findings suggest that by optimizing different integral performance criteria using this design technique, controllers for both integer and fractional-order plants can yield favorable step responses. The proposed algorithm is validated by comparing its step response performance with that of previous research, followed by a discussion on sensitivity analysis and computational requirements.

关键词： Optimization Uncertainty Fuzzy logic Power system stability Fourier series approximation algorithms Transfer functions Fourier series method fractional-order PID controller type-2 fuzzy controller

来源：评论

学校读者我要写书评

暂无评论

SHATTER: Searching Heterogeneous Combat Network Attack Sequences Through Network Embedding and Reinforcement Learning

引用

IEEE SYSTEMS JOURNAL 2023年第3期17卷 4497-4508页

作者： Zeng, Chengyi Liu, Hongfu Lu, Lina Chen, Jing Zhou, Zongtan Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha Peoples R China Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha Peoples R China Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha Peoples R China Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha Peoples R China Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha Peoples R China

A distributed combat system consists of various types of combat units interconnected through the network, which can be abstracted as a complex heterogeneous combat network (HCN). It is of great military value to study the disintegration of the operational capability of HCNs for operational mission planning and improving the survivability of operational networks. Heterogeneous networks can effectively fuse more information than homogeneous networks can. The function implementation of heterogeneous networks is also more complex. Finding the key nodes and obtaining an efficient attack sequence is our focus. Most current research on network disintegration has difficulty achieving efficiency and timeliness. In this article, we specifically propose searching heterogeneous combat network attack sequences through network embedding and reinforcement learning (SHATTER) model, which uses the induction algorithm combined with the graph neural network and reinforcement learning to solve the network disintegration sequence problem. Through experiments with synthetic heterogeneous network datasets and real heterogeneous networks, we determined that SHATTER significantly outperformed existing technologies.

关键词： Reinforcement learning Heterogeneous networks Deep learning Task analysis approximation algorithms Signal processing algorithms Training Graph neural network (GNN) heterogeneous combat network (HCN) network disintegration reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Mixed-Precision Sparse Approximate Inverse Preconditioning Algorithm on GPU

引用

IEEE ACCESS 2023年 11卷 136410-136421页

作者： Chu, Xinyue Nanjing Normal Univ Sch Comp & Elect Informat Nanjing 210023 Peoples R China

In this study, in order to further improve the construction efficiency of sparse approximate inverse (SPAI) preconditioners, we attempt to explore the construction method of SPAI preconditioners in mixed-precision mode from the perspective of single and double precision mixing, and thus propose two mixed-precision SPAI preconditioning algorithms on GPU, abbreviated as MP-SSPAI and MP-HeuriSPAI, respectively. In MP-SSPAI, with original static SPAI preconditioning algorithm as the research object, we mainly consider the following factors to construct its preconditioner in mixed-precision mode: 1) use single precision as much as possible to improve computational efficiency of the preconditioner while ensuring its validity;2) store certain components in single precision after they have been determined to require single-precision computation to improve read efficiency;and 3) maintain the high-precision output of the preconditioner to ensure that it is computed with high precision when applied to the iterative algorithm. In MP-HeuriSPAI, a mixed-precision heuristic dynamic SPAI preconditioning algorithm on GPU is presented based on the above factors, using HeuriSPAI as the object of study. The experimental results demonstrate the effectiveness and high performance of the proposed MP-SSPAI and MP-HeuriSPAI by comparing them with their respective double-precision versions, single-precision versions, and extended versions.

关键词： Heuristic algorithms Matrix decomposition approximation algorithms Graphics processing units Iterative algorithms Sparse matrices Minimization GPU mixed precision preconditioning algorithm sparse approximate inverse

来源：评论

学校读者我要写书评

暂无评论

Network Topology Optimization via Deep Reinforcement Learning

引用

IEEE TRANSACTIONS ON COMMUNICATIONS 2023年第5期71卷 2847-2859页

作者： Li, Zhuoran Wang, Xing Pan, Ling Zhu, Lin Wang, Zhendong Feng, Junlan Deng, Chao Huang, Longbo Tsinghua Univ Inst Interdisciplinary Informat Sci IIIS Beijing 100084 Peoples R China China Mobile Res Inst Beijing 100053 Peoples R China

Topology impacts important network performance metrics, including link utilization, throughput and latency, and is of central importance to network operators. However, due to the combinatorial nature of network topology, it is extremely difficult to obtain an optimal solution, especially since topology planning in networks also often comes with management-specific constraints. As a result, local optimization with hand-tuned heuristic methods from human experts is often adopted in practice. Yet, heuristic methods cannot cover the global topology design space while taking into account constraints, and cannot guarantee to find good solutions. In this paper, we propose a novel deep reinforcement learning (DRL) algorithm for graph searching, called DRL-GS, for network topology optimization. DRL-GS consists of three novel components, including a verifier to validate the correctness of a generated network topology, a graph neural network (GNN) to efficiently approximate topology rating, and a DRL agent to conduct a topology search. DRL-GS can efficiently search over relatively large topology space and output topology with satisfactory performance. We conduct a case study based on a real-world network scenario, and our experimental results demonstrate the superior performance of DRL-GS in terms of both efficiency and performance.

关键词： Network topology Topology Optimization approximation algorithms Graph neural networks Costs Planning nonlinear combinatorial optimization deep reinforcement learning graph neural network

来源：评论

学校读者我要写书评

暂无评论

On Performance and Calibration of Natural Gradient Langevin Dynamics

引用

IEEE ACCESS 2023年 11卷 53919-53931页

作者： Robbani, Hanif Amal Bustamam, Alhadi Adnan, Risman Ahmad, Shandar Univ Indonesia Dept Math Depok 16424 Indonesia Kalbe Digital Lab Jakarta 10510 Indonesia Jawaharlal Nehru Univ Sch Computat & Integrat Sci Delhi 110067 Delhi India

Producing deep neural network (DNN) models with calibrated confidence is essential for applications in many fields, such as medical image analysis, natural language processing, and robotics. Modern neural networks have been reported to be poorly calibrated compared with those from a decade ago. The stochastic gradient Langevin dynamics (SGLD) algorithm offers a tractable approximate Bayesian inference applicable to DNN, providing a principled method for learning the uncertainty. A recent benchmark study showed that SGLD could produce a more robust model to covariate shifts than other competing methods. However, vanilla SGLD is also known to be slow, and preconditioning can improve SGLD efficacy. This paper proposes eigenvalue-corrected Kronecker factorization (EKFAC) preconditioned SGLD (EKSGLD), in which a novel second-order gradient approximation is employed as a preconditioner for the SGLD algorithm. This approach is expected to bring together the advantages of both second-order optimization and the approximate Bayesian method. Experiments were conducted to compare the performance of EKSGLD with existing preconditioning methods and showed that it could achieve higher predictive accuracy and better calibration on the validation set. EKSGLD improved the best accuracy by 3.06% on CIFAR-10 and 4.15% on MNIST, improved the best negative log-likelihood by 16.2% on CIFAR-10 and 11.4% on MNIST, and improved the best thresholded adaptive calibration error by 4.05% on CIFAR-10.

关键词： Calibration Benchmark testing Bayes methods Uncertainty Deep learning approximation algorithms Optimization Predictive methods Natural gradient second-order optimization Bayesian deep learning Langevin dynamics confidence calibration predictive uncertainty

来源：评论

学校读者我要写书评

暂无评论

Internal Boundary Between Entanglement and Separability Within a Quantum State

引用

IEEE TRANSACTIONS ON INFORMATION THEORY 2023年第1期69卷 251-261页

作者： Wang, Bang-Hai Guangdong Univ Technol Sch Comp Sci & Technol Guangzhou 510006 Peoples R China Univ Oxford Dept Phys Oxford OX1 3PU England

Quantum states are the key mathematical objects in quantum mechanics, and entanglement lies at the heart of the nascent fields of quantum information processing and computation. However, there has not been a general, necessary and sufficient, and operational separability condition to determine whether an arbitrary quantum state is entangled or separable. In this paper, we show that whether a quantum state is entangled or not is determined by a threshold within the quantum state. We first introduce the concept of finer and optimal separable states based on the properties of separable states in the role of higher-level witnesses. Then we show that any bipartite quantum state can be decomposed into a convex mixture of its optimal entangled state and its optimal separable state. Furthermore, we show that whether an arbitrary quantum state is entangled or separable, as well as positive partial transposition (PPT) or not, is determined by the robustness of its optimal entangled state to its optimal separable state with reference to a crucial threshold. Moreover, for an arbitrary quantum state, we provide operational algorithms to obtain its optimal entangled state, its optimal separable state, its best separable approximation (BSA) decomposition, and its best PPT approximation decomposition. And the open question of how to calculate the BSA in high-dimension systems is partially done.

关键词： Quantum entanglement Quantum state Quantum mechanics Quantum computing Heart approximation algorithms Teleportation Best separable approximation crucial threshold internal structure operational algorithms quantum entanglement

来源：评论

学校读者我要写书评

暂无评论

A Sketch Framework for Approximate Data Stream Processing in Sliding Windows

引用

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 2023年第5期35卷 4411-4424页

作者： Gou, Xiangyang Zhang, Yinda Hu, Zhoujing He, Long Wang, Ke Liu, Xilai Yang, Tong Wang, Yi Cui, Bin Peking Univ Sch Comp Sci Beijing 100871 Peoples R China Peking Univ Natl Engn Lab Big Data Anal Technol & Applicat Beijing 100871 Peoples R China Southern Univ Sci & Technol Inst Future Networks Shenzhen 518055 Peoples R China Peng Cheng Lab Shenzhen 518066 Peoples R China

Data stream processing has become a hot issue in recent years. There are three fundamental stream processing tasks: membership query, frequency query, and Top-K query. While most existing solutions address these queries in fixed windows, this paper focuses on a more challenging task: answering these queries in sliding windows. While most existing solutions address different kinds of queries by using different algorithms, this paper focuses on a generic framework. In this paper, we propose a generic framework, namely the Sliding sketch, which can be applied to many existing solutions for the above three queries, and enable them to support queries in sliding windows. We apply our framework to five state-of-the-art sketches for the above three kinds of queries. Theoretical analysis and extensive experimental results show that the accuracy of existing sketches that do not support sliding windows becomes much higher than the corresponding prior art after using our framework. We released all the source code at Github.

关键词： Probabilistic logic Hash functions Data models Computer science Big Data Art approximation algorithms Data stream sliding window sketch approximate query

来源：评论

学校读者我要写书评

暂无评论

Massive MIMO Detectors Based on Deep Learning, Stair Matrix, and Approximate Matrix Inversion Methods

引用

IEEE ACCESS 2023年 11卷 100268-100283页

作者： Albreem, Mahmoud A. Alnajjar, Khawla A. Almasadeh, Ali J. Univ Sharjah Dept Elect Engn Sharjah U Arab Emirates

Massive multiple-input multiple-output (MIMO) is an essential technology in fifth-generation (5G) and beyond 5G (B5G) communication systems. Massive MIMO is employed to meet the increasing request for high capacity in next-generation wireless communication networks. However, signal processing in massive MIMO incurs a high complexity due to a large number of transmitting and receiving antenna elements. In this paper, we propose low complexity massive MIMO data detection techniques based on zero-forcing (ZF) and vertical bell laboratories layered space-time (V-BLAST) method in combination with approximate matrix inversion techniques;Neumann series (NS) and Newton iteration (NI). The proposed techniques reduce the complexity of the ZF V-BLAST method since they avoid the exact matrix inverse computation. Initialization based on a stair matrix is also exploited to balance the performance and the complexity. In addition, we propose a massive MIMO detector based on approximate matrix inversion with a stair matrix initialization and deep learning (DL) based detector;MM Network (MMNet) algorithm. MMNet contains a linear transformation followed by a non-linear denoising stage. As signals propagate through the MMNet, the noise distribution at the input of the denoiser stages approaches a Gaussian distribution, form precisely the conditions in which the denoisers can attenuate noise maximally. We validated the performance of the proposed massive MIMO detection schemes in Gaussian and realistic channel models, i.e., Quadriga channels models. Simulations demonstrate that the proposed detectors achieve a remarkable improvement in the performance with a notable computational complexity reduction when compared to conventional ZF V-BLAST and the MMNET in both simple and real channel scenarios.

关键词： Detectors Massive MIMO Iterative methods Computational complexity approximation algorithms Modulation 5G mobile communication MIMO communication Deep learning Data models 5G massive MIMO Gauss-Seidel successive overrelaxation Neumann series Newton iteration deep learning data detection stair matrix diagonal matrix V-BLAST

来源：评论

学校读者我要写书评

暂无评论

Analytical Algorithm for Capacities of Classical and Classical-Quantum Channels

引用

IEEE TRANSACTIONS ON INFORMATION THEORY 2023年第3期69卷 1381-1393页

作者： Hayashi, Masahito Southern Univ Sci & Technol Shenzhen Inst Quantum Sci & Engn Shenzhen Nanshan Peoples R China

We derive an analytical algorithm for the channel capacity of a classical channel without any iteration, while its existing algorithms require iterations and the number of iterations depends on the required precision level. Hence, our algorithm is its first analytical algorithm for this task without any iteration, while this algorithm needs several conditions for the channel. We apply the obtained algorithm to examples, and see how the obtained algorithm works in these examples. Then, we extend it to the channel capacity of a classical-quantum (cq-) channel. Many existing studies proposed algorithms for a cq-channel and all of them require iterations. Our extended analytical algorithm has also no iteration, and outputs the exactly optimum value.

关键词： Channel capacity approximation algorithms STEM Mathematical models Complexity theory Iterative algorithms Upper bound Mutual information maximization channel capacity classical-quantum channel analytical algorithm

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：