检索结果-内蒙古大学图书馆

Convolutional Neural Networks-Based Lung Nodule Classification: A Surrogate-Assisted evolutionary algorithm for Hyperparameter Optimization

引用

IEEE TRANSACTIONS ON evolutionary COMPUTATION 2021年第5期25卷 869-882页

作者： Zhang, Miao Li, Huiqi Pan, Shirui Lyu, Juan Ling, Steve Su, Steven Beijing Inst Technol Sch Informat & Elect Beijing 100081 Peoples R China Univ Technol Sydney Fac Engn & Informat Technol Ultimo NSW 2007 Australia Monash Univ Fac Informat Technol Melbourne Vic 3800 Australia Harbin Engn Univ Informat & Commun Engn Harbin 150001 Peoples R China

This article investigates deep neural networks (DNNs)-based lung nodule classification with hyperparameter optimization. Hyperparameter optimization in DNNs is a computationally expensive problem, and a surrogate-assisted evolutionary algorithm has been recently introduced to automatically search for optimal hyperparameter configurations of DNNs, by applying computationally efficient surrogate models to approximate the validation error function of hyperparameter configurations. Different from existing surrogate models adopting stationary covariance functions (kernels) to measure the difference between hyperparameter points, this article proposes a nonstationary kernel that allows the surrogate model to adapt to functions whose smoothness varies with the spatial location of inputs. A multilevel convolutional neural network (ML-CNN) is built for lung nodule classification, and the hyperparameter configuration is optimized by the proposed nonstationary kernel-based Gaussian surrogate model. Our algorithm searches with a surrogate for optimal setting via a hyperparameter importance-based evolutionary strategy, and the experiments demonstrate our algorithm outperforms manual tuning and several well-established hyperparameter optimization methods, including random search, grid search, the tree-structured parzen estimator (TPE) approach, Gaussian processes (GP) with stationary kernels, and the recently proposed hyperparameter optimization via RBF and dynamic (HORD) coordinate search.

关键词： Optimization Kernel Neural networks Lung Feature extraction Gaussian processes Convolutional neural networks AutoML evolutionary algorithm hyperparameter optimization lung nodule classification nonstationary kernel

来源：评论

学校读者我要写书评

暂无评论

A proper infill sampling strategy for improving the speed performance of a Surrogate-Assisted evolutionary algorithm

引用

COMPUTERS & STRUCTURES 2017年 178卷 58-70页

作者： Vincenzi, Loris Gambarelli, Paola Univ Modena & Reggio Emilia DIEF Via Vivarelli 10 I-41135 Modena Italy

In the present paper, an improved Surrogate-Assisted evolutionary algorithm is proposed. It combines the Differential Evolution algorithm with a quadratic surrogate approximation and a proper infill sampling strategy to choose appropriate sample points. The selection of the new candidate points is arranged to enhance both the local accuracy and the global optimum search. A comparison between performances of different evolutionary algorithms is carried out by searching the global minimum of two benchmark functions, by solving a dynamic identification problem of a three floor frame and by calibrating the non-linear stress-crack opening relation for Fibre-Reinforced Concrete specimens starting from experimental data. (C) 2016 Elsevier Ltd. All rights reserved.

关键词： Surrogate model evolutionary algorithm Model updating Differential evolution Inn sampling strategy

来源：评论

学校读者我要写书评

暂无评论

An adaptive clustering-based evolutionary algorithm for many-objective optimization problems

引用

INFORMATION SCIENCES 2020年 537卷 261-283页

作者： Liu, Songbai Yu, Qiyuan Lin, Qiuzhen Tan, Kay Chen Shenzhen Univ Coll Comp Sci & Software Engn Shenzhen Peoples R China City Univ Hong Kong Dept Comp Sci Hong Kong Peoples R China

This paper proposes an adaptive clustering-based evolutionary algorithm for many-objective optimization problems (MaOPs), called MaOEA/AC. In this algorithm, an adaptive clustering strategy (ACS) is first introduced to divide the population into multiple clusters, which can properly fit various Pareto fronts (PFs) of the target MaOPs. Then, the environmental selection of MaOEA/AC is designed based on these clusters to collect the solutions with balanceable convergence and diversity. To be more detail, the similarity between solutions in ACS is appropriately measured by computing the Euclidean distance between their projections on an adaptive unit hyper-surface, whose curving rate is controlled by a parameter p. A simple yet effective estimation method is proposed to get a suitable value of p based on the distribution of the current non-dominated solution set, so that the estimated unit hyper-surface can roughly reflect the characteristics of PFs in the target MaOPs. The effectiveness of MaOEA/AC is validated by numerous experimental studies on solving test MaOPs with various PFs, which have the characteristics with convex, concave, inverted, disconnected, degenerated, and other mixed or irregular PFs. The experiments also show that MaOEA/AC has the superior performance over several recent many-objective evolutionary algorithms, when solving most of these test MaOPs. (C) 2020 Elsevier Inc. All rights reserved.

关键词： evolutionary algorithm Many-objective optimization Adaptive clustering

来源：评论

学校读者我要写书评

暂无评论

Two-Archive evolutionary algorithm Based on Multi-Search Strategy for Many-Objective Optimization

引用

IEEE ACCESS 2019年 7卷 79277-79286页

作者： Dai, Cai Shaanxi Normal Univ Sch Comp Sci Xian 710119 Shaanxi Peoples R China

Taking both convergence and diversity into consideration, this paper proposes a two-archive an evolutionary algorithm based on multi-search strategy (TwoArchM) to cope with many-objective optimization problems. The basic idea is to use two separate archives to balance the convergence and diversity and use a multi-search strategy to improve convergence and diversity. To be specific, two updated strategies are adopted to maintain diversity and improve the convergence, respectively;a multi-search strategy is utilized to balance exploration and exploitation. A search strategy selects convergent solutions from offspring and two archives as parents to enhance the convergence;the goal of another search strategy is to balance exploration and exploitation. The TwoArchM is compared experimentally with several state-of-the-art algorithms on the CEC2018 many-objective benchmark functions with up to 15 objectives and the experimental results verify the competitiveness and effectiveness of the proposed algorithm.

关键词： Many-objective optimization two archives multi-search strategy evolutionary algorithm

来源：评论

学校读者我要写书评

暂无评论

An Indicator-Based Many-Objective evolutionary algorithm With Boundary Protection

引用

IEEE TRANSACTIONS ON CYBERNETICS 2021年第9期51卷 4553-4566页

作者： Liang, Zhengping Luo, Tingting Hu, Kaifeng Ma, Xiaoliang Zhu, Zexuan Shenzhen Univ Coll Comp Sci & Software Engn Shenzhen 518060 Peoples R China Shenzhen Pengcheng Lab Shenzhen 518055 Peoples R China Shenzhen Univ Shenzhen Inst Artificial Intelligence & Robot Soc SZU Branch Shenzhen 518060 Peoples R China

Many-objective optimization problems (MaOPs) pose a big challenge to the traditional Pareto-based multiobjective evolutionary algorithms (MOEAs). As the number of objectives increases, the number of mutually nondominated solutions explodes and MOEAs become invalid due to the loss of Pareto-based selection pressure. Indicator-based many-objective evolutionary algorithms (MaOEAs) have been proposed to address this issue by enhancing the environmental selection. Indicator-based MaOEAs are easy to implement and of good versatility, however, they are unlikely to maintain the population diversity and coverage very well. In this article, a new indicator-based MaOEA with boundary protection, namely, MaOEA-IBP, is presented to relieve this weakness. In MaOEA-IBP, a worst elimination mechanism based on the I-epsilon+ indicator and boundary protection strategy is devised to enhance the balance of population convergence, diversity, and coverage. Specifically, a pair of solutions with the smallest I-epsilon+ value are first identified from the population. If one solution dominates the other, the dominated solution is eliminated. Otherwise, one solution is eliminated by the boundary protection strategy. MaOEA-IBP is compared with four indicator-based algorithms (i.e., I-SDE+ , SRA, MaOEAIGD, and ARMOEA) and other five state-of-the-art MaOEAs (i.e., KnEA, MaOEA-CSS, 1by1EA, RVEA, and EFR-RR) on various benchmark MaOPs. The experimental results demonstrate that MaOEA-IBP can achieve competitive performance with the compared algorithms.

关键词： I-epsilon+ indicator boundary protection convergence coverage diversity evolutionary algorithm many-objective evolutionary algorithm (MaOEA) multiobjective evolutionary algorithm (MOEA)

来源：评论

学校读者我要写书评

暂无评论

Neural Network Ensemble With evolutionary algorithm for Highly Imbalanced Classification

引用

IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTATIONAL INTELLIGENCE 2023年第5期7卷 1394-1404页

作者： Sun, Poly Z. H. Zuo, Tian-Yu Law, Rob Wu, Edmond Q. Song, Aiguo Shanghai Jiao Tong Univ Sch Mech Engn Dept Ind Engn Shanghai 200240 Peoples R China Nanjing Univ Informat Sci & Technol Sch Automat Nanjing 210044 Peoples R China Univ Macau Asia Pacific Acad Econ & Management Fac Business Adm Dept Integrated Resort & Tourism Management Macau 0999078 Peoples R China Shanghai Jiao Tong Univ Dept Automat Shanghai 200240 Peoples R China Southeast Univ Sch Instrument Sci & Engn Nanjing 210096 Peoples R China

Imbalanced data is a major challenge in classification tasks. Most classification algorithms tend to be biased toward the samples in the majority class but fail to classify the samples in the minority class. Recently, ensemble learning, as a promising method, has been rapidly developed in solving highly imbalanced classification. However, the design of the base classifier for the ensemble is still an open question because the optimization problem of the base classifier is gradientless. In this study, the evolutionary algorithm (EA) technique is adopted to solve a wide range of optimization design problems in highly imbalanced classification without gradient information. A novel EA-based classifier optimization design method is proposed to optimize the design of multiple base classifiers automatically for the ensemble. In particular, an EA method with a neural network (NN) as the base classifier termed NN ensemble with EA (NNEAE) is developed for highly imbalanced classification. To verify the performance of NNEAE, extensive experiments are designed for testing. Results illustrate that NNEAE outperforms other compared methods.

关键词： Artificial neural networks Training Ensemble learning Costs Bagging Support vector machines Sun Network architecture search evolutionary algorithm classification imbalanced data

来源：评论

学校读者我要写书评

暂无评论

Optimization of a Darrieus vertical-axis wind turbine using blade element - momentum theory and evolutionary algorithm

引用

RENEWABLE ENERGY 2013年 59卷 184-192页

作者： Bedon, Gabriele Castelli, Marco Raciti Benini, Ernesto Univ Padua Dept Ind Engn I-35131 Padua Italy

Wind turbine design procedures usually involve the adoption of the blade element - momentum theory. Nevertheless, its use is limited by the lack of extended database regarding the aerodynamic coefficients for most used airfoils. In the present work, an extended database generation procedure for symmetric profiles is discussed and validated with the aim of adopting numerical optimization methods for vertical-axis wind turbine design. evolutionary algorithms are thereby utilized to provide optimal configurations for different design objectives. The pure performance and the annual energy production are here considered in order to show the capabilities of the numerical code. A relevant increase in performance is achieved for all the obtained results, showing that the numerical optimization can be successfully adopted in vertical-axis wind turbine design procedures. (C) 2013 Elsevier Ltd. All rights reserved.

关键词： Wind turbine Blade element - momentum (BE-M) theory Extended airfoil database evolutionary algorithm Numerical optimization WOMBAT

来源：评论

学校读者我要写书评

暂无评论

Vibrational spectra of Ruthenium Carbide structures yielded by the structure search employing evolutionary algorithm

引用

MATERIALS SCIENCE IN SEMICONDUCTOR PROCESSING 2015年 40卷 484-490页

作者： Harikrishnan, G. Ajith, K. M. Chandra, Sharat Valsakumar, M. C. Natl Inst Technol Karnataka Dept Phys Surathkal 575025 Karnataka India Indira Gandhi Ctr Atom Res Div Mat Sci Kalpakkam 603102 Tamil Nadu India Hyderabad Cent Univ Sch Engn Sci & Technol Telangana 500046 India

Out of the three dynamically stable structures of Ruthenium Carbides yielded by the exhaustive structure search employing evolutionary algorithm, Born effective charges are computed for the semiconducting RuC in Zinc blende structure using density functional perturbation theory. Using the phonon frequencies and the Born effective charge tensors of Ru and C in this structure, infrared spectrum is generated for this system. Computations of these dynamical quantities and IR spectra from first principles can be helpful in the unambiguous determination of the stoichiometry and structure by comparison of the experimental measurements with the computational predictions. The positive formation energies of the three systems show that high pressure and possibly high temperature may be necessary for their synthesis. Formation energies of these systems at different pressures are computed. One of the structurally stable systems, Ru3C with hexagonal structure (P (6) over bar m2), has negative formation energy at 200 GPa. The system reported from the first synthesis of Ruthenium Carbide also has the same symmetry, though it has a different stoichiometry. (C) 2015 Elsevier Ltd. All rights reserved.

关键词： Ruthenium Carbides Density functional theory evolutionary algorithm Born effective charges IR spectra

来源：评论

学校读者我要写书评

暂无评论

An Adaptive Two-Population evolutionary algorithm for Constrained Multi-Objective Optimization Problems

引用

IEEE ACCESS 2023年 11卷 82118-82131页

作者： Zhao, Kaiwen Wang, Peng Tong, Xiangrong Yantai Univ Sch Comp & Control Engn Yantai 264005 Peoples R China

Striking a balance between objective optimization and constraint satisfaction is essential for solving constrained multi-objective optimization problems (CMOPs). Nevertheless, most existing evolutionary algorithms face significant challenges on CMOPs with intricate infeasible regions. To tackle these challenges, this paper proposes an adaptive two-population evolutionary algorithm, named ATEA, which dynamically exploits promising information under infeasible solutions to facilitate objective optimization and constraint satisfaction. Specifically, a two-population collaboration mechanism is designed to balance the unconstrained Pareto front search and constrained Pareto front search. Moreover, an adaptive constraint handling strategy is presented to reasonably deploy search resources. Furthermore, a promising infeasibility-based environmental selection and an elitist feasibility-based environmental selection are developed for the two populations to break through complex infeasible barriers and enhance selection pressure, respectively. Comparison experimental results of ATEA with five state-of-the-art algorithms on 33 benchmark test problems and 4 real-word CMOPs demonstrate that ATEA performs competitively with the chosen designs.

关键词： Constrained multi-objective optimization evolutionary algorithm two-population collaboration mechanism multi-state objective optimization constraint satisfaction

来源：评论

学校读者我要写书评

暂无评论

Multiobjective optimization using coupled response surface model and evolutionary algorithm

引用

AIAA JOURNAL 2005年第6期43卷 1316-1325页

作者： Lian, YS Liou, MS Ohio Aerosp Inst Cleveland OH 44142 USA NASA John H Glenn Res Ctr Lewis Field Cleveland OH 44135 USA

In this work we develop an efficient approach for computationally expensive multiobjective design optimization problems. In this approach we bring together design of experiment, a response surface model, a genetic algorithm, and computational-fluid-dynamics analysis tools to provide an integrated optimization system. We use an improved hypercube sampling to preselect an array of design points on which the computational-fluid-dynamics code will run. Then a computationally cheap surrogate model is constructed based on response surface approximation. A real-coded genetic algorithm is then applied on the surrogate model to perform multiobjective optimization. Representative solutions are chosen from the Pareto-optimal front to verify against the computational-fluid-dynamics code. This proposed method is used in the redesign of a single-stage turbopump, a two-stage turbopump, and the NASA rotor67 transonic compressor blade. For the single-stage pump optimization problem, we can improve the total head rise by 1.2% with the same power input;for the multistage pump problem, we can improve the total head rise by 0.5% at the same power input;for the rotor67 compressor blade design, we can increase the pressure ratio by 1.8% or reduce the entropy generation by 6.2%. We achieve these with a much reduced computational cost.

关键词： Surface Modeling Turbopumps Genetic algorithm Surrogate Model CFD Transonic Compressors Central Composite Designs Entropy Generation evolutionary algorithm Computational Fluid Dynamics Code

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：