检索结果-内蒙古大学图书馆

Greedy active learning algorithm for logistic regression models

COMPUTATIONAL STATISTICS & DATA ANALYSIS 2019年 129卷 119-134页

作者： Hsu, Hsiang-Ling Chang, Yuan-chin Ivan Chen, Ray-Bing Natl Univ Kaohsiung Kaohsiung Taiwan Acad Sinica Taipei Taiwan Natl Cheng Kung Univ Tainan Taiwan

We study a logistic model-based active learning procedure for binary classification problems, in which we adopt a batch subject selection strategy with a modified sequential experimental design method. Moreover, accompanying the proposed subject selection scheme, we simultaneously conduct a greedy variable selection procedure such that we can update the classification model with all labeled training subjects. The proposed algorithm repeatedly performs both subject and variable selection steps until a prefixed stopping criterion is reached. Our numerical results show that the proposed procedure has competitive performance, with smaller training size and a more compact model compared with that of the classifier trained with all variables and a full data set. We also apply the proposed procedure to a well-known wave data set (Breiman et al., 1984) and a MAGIC gamma telescope data set to confirm the performance of our method, (C) 2018 Elsevier B.V. All rights reserved.

关键词： active learning algorithm D-efficiency criterion Forward selection Graft optimization

来源：评论

学校读者我要写书评

暂无评论

Efficient reliability analysis of slopes in spatially variable soils with active learning-assisted bootstrap polynomial chaos expansion

引用

COMPUTERS AND GEOTECHNICS 2025年 179卷

作者： Liao, Kang Zhao, Xiaoyan Wu, Yiping Miao, Fasheng Pan, Yutao Beer, Michael Southwest Jiaotong Univ Fac Geosci & Engn Chengdu 610031 Peoples R China China Univ Geosci Fac Engn Wuhan 430074 Peoples R China Norwegian Univ Sci & Technol Dept Civil & Environm Engn N-7047 Trondheim Norway Leibniz Univ Hannover Inst Risk & Reliabil D-30167 Hannover Germany

Evaluating the reliability of slopes with spatial variability is a challenging issue, especially when the failure probably of the target event is at a low level, because of unaffordable computational cost required in such cases. In this context, an adaptive surrogate model-based approach, namely active learning-assisted bootstrap polynomial chaos expansion, is proposed to alleviate the above computational burden. The proposed approach extends the traditional polynomial chaos expansion by introducing the bootstrap resampling method so that it can deal with reliability issues smoothly and provide a feasible configuration environment to support the active learning algorithm. The computational efficiency can thus be greatly improved by adaptively searching for the most informative samples to train the surrogate model through iterative program. Two spatially varying soil slopes are studied to illustrate the validity of the active learning-assisted bootstrap polynomial chaos expansion. The results show that the proposed approach has superior advantages in terms of efficiency and accuracy, and it is also suitable for handling problems with complex parameter configurations, including high dimensionality and cross-correlation. Besides, the proposed approach has potential in addressing geotechnical engineering problems with low probability levels.

关键词： Spatial variability Surrogate model active learning algorithm Bootstrap polynomial chaos expansion

来源：评论

学校读者我要写书评

暂无评论

active learning combining uncertainty and diversity for multi-class image classification

引用

IET COMPUTER VISION 2015年第3期9卷 400-407页

作者： Gu, Yingjie Jin, Zhong Chiu, Steve C. Nanjing Univ Sci & Technol Dept Comp Sci & Engn Nanjing 210094 Jiangsu Peoples R China Idaho State Univ Dept Elect Engn Pocatello ID 83209 USA

In computer vision and pattern recognition applications, there are usually a vast number of unlabelled data whereas the labelled data are very limited. active learning is a kind of method that selects the most representative or informative examples for labelling and training;thus, the best prediction accuracy can be achieved. A novel active learning algorithm is proposed here based on one-versus-one strategy support vector machine (SVM) to solve multi-class image classification. A new uncertainty measure is proposed based on some binary SVM classifiers and some of the most uncertain examples are selected from SVM output. To ensure that the selected examples are diverse from each other, Gaussian kernel is adopted to measure the similarity between any two examples. From the previous selected examples, a batch of diverse and uncertain examples are selected by the dynamic programming method for labelling. The experimental results on two datasets demonstrate the effectiveness of the proposed algorithm.

关键词： image classification computer vision learning (artificial intelligence) support vector machines dynamic programming multiclass image classification computer vision pattern recognition applications unlabelled data active learning algorithm support vector machine binary SVM classiflers Gaussian kernel dynamic programming method

来源：评论

学校读者我要写书评

暂无评论

active learning on stacked machine learning techniques for predicting compressive strength of alkali-activated ultra-high-performance concrete

引用

ARCHIVES OF CIVIL AND MECHANICAL ENGINEERING 2024年第1期25卷 1-35页

作者： Kazemi, Farzin Shafighfard, Torkan Jankowski, Robert Yoo, Doo-Yeol Gdansk Univ Technol Fac Civil & Environm Engn Ul G Narutowicza 11-12 PL-80233 Gdansk Poland Univ Naples Federico II Sch Polytech & Basic Sci Dept Struct Engn & Architecture I-80125 Naples Italy Polish Acad Sci Inst Fluid Flow Machinery Generala Jozefa Fiszera 14 PL-80231 Gdansk Poland Yonsei Univ Dept Architecture & Architectural Engn 50 Yonsei Ro Seoul 03722 South Korea

Conventional ultra-high performance concrete (UHPC) has excellent development potential. However, a significant quantity of CO2 is produced throughout the cement-making process, which is in contrary to the current worldwide trend of lowering emissions and conserving energy, thus restricting the further advancement of UHPC. Considering climate change and sustainability concerns, cementless, eco-friendly, alkali-activated UHPC (AA-UHPC) materials have recently received considerable attention. Following the emergence of advanced prediction techniques aimed at reducing experimental tools and labor costs, this study provides a comparative study of different methods based on machine learning (ML) algorithms to propose an active learning-based ML model (AL-Stacked ML) for predicting the compressive strength of AA-UHPC. A data-rich framework containing 284 experimental datasets and 18 input parameters was collected. A comprehensive evaluation of the significance of input features that may affect compressive strength of AA-UHPC was performed. Results confirm that AL-Stacked ML-3 with accuracy of 98.9% can be used for different general experimental specimens, which have been tested in this research. active learning can improve the accuracy up to 4.1% and further enhance the Stacked ML models. In addition, graphical user interface (GUI) was introduced and validated by experimental tests to facilitate comparable prospective studies and predictions.

关键词： Machine learning algorithm Alkali-activated ultra-high performance concrete active learning algorithm Compressive strength Hyperparameter optimization

来源：评论

学校读者我要写书评

暂无评论

An optimization-based method of calibrating load and resistance factors: Application to slope and breakwaters' foundation stability

引用

OCEAN ENGINEERING 2024年第Part1期313卷

作者： Doan, Nhu Son Mac, Van Ha Dinh, Huu-Ba Vietnam Maritime Univ Fac Civil Engn 484 Lach Tray Str Haiphong City Vietnam Univ Transport & Commun Fac Civil Engn Dept Urban Transport & Marine Coastal Engn 3 Cau Giay St Hanoi Vietnam Van Lang Univ Inst Computat Sci & Artificial Intelligence Lab Computat Civil Engn Ho Chi Minh City Vietnam Van Lang Univ Fac Civil Engn Sch Technol Ho Chi Minh City Vietnam

Monte Carlo simulation (MCS)-based calibrations can accurately determine probabilistic load and resistance factors (LRFs) needed in the limit state designs. However, most of the computing time and effort of the calibrations is for evaluating performance functions if they are defined implicitly, as in the case of slope stabilities. This study proposes a robust framework that combines the advantages of adaptive artificial neural networks (ANNs) in approximating implicit performance functions with an optimization process to establish the LRFs quickly and automatically. Furthermore, experiment data obtained in the preceding iterations of the optimizations are accumulated and reused in the subsequent trial. For illustration, three implicit problems, including a slope and two cases of breakwater foundation stability, are examined to demonstrate the efficiency and accuracy of the proposed procedure. The investigations show that the proposed framework can be accurately completed within 1 h instead of lasting for weeks of calculation when using basic MCSs. Remarkably, reusing experiment data helps decrease the necessary data by two-thirds compared to only using the adaptive ANN, facilitating a faster calibration process. Thus, this work contributes a practical method for calibrating LRFs needed in limit state designs of geotechnical engineering fields wherein limit state equations are defined in implicit fashions.

关键词： Load and resistance factors calibration MCS-based reliability analysis Slope stability active learning algorithm Reliability-based optimization

来源：评论

学校读者我要写书评

暂无评论

Multi-objective aerodynamic optimization using active multi-output Gaussian process and mesh deformation method

引用

PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS PART G-JOURNAL OF AEROSPACE ENGINEERING 2022年第4期236卷 767-776页

作者： Yang, Fan Chen, Zhaolin Nanjing Univ Aeronaut & Astronaut State Key Lab Mech & Control Mech Struct Nanjing Peoples R China Nanjing Univ Aeronaut & Astronaut Coll Aerosp Engn Nanjing Peoples R China Xi An Jiao Tong Univ State Key Lab Strength & Vibrat Mech Struct Xian Peoples R China

A wing is an important part of the aircraft to improve aerodynamic performance. The current study is focused on an adaptive surrogate algorithm for airfoil aerodynamic optimization, which is based on a multi-output Gaussian process model. The conventional design method seriously relies on wind tunnel experiments and expensive computational simulations. The metamodels can significantly improve design efficiency and hence reduce the overall design costs. An active learning algorithm is proposed to improve the effectiveness of the multi-output Gaussian process model. The NSGA-II algorithm is adopted to obtain the optimal Pareto set with the optimization objectives of lift and drag coefficients for adaptive airfoil shapes. Besides, the Bezier curve and radial basis function are utilized in this study for airfoil mesh deformation. The results show that the airfoil shape can be obtained effectively by integrating the metamodel, active learning algorithm, and multi-objective optimization algorithm. The optimized results are of great engineering applications.

关键词： Aerodynamic shape design mesh deformation metamodel active learning algorithm multi-objective optimization

来源：评论

学校读者我要写书评

暂无评论

Efficient computing technique for reliability analysis of high-dimensional and low-failure probability problems using active learning method

引用

PROBABILISTIC ENGINEERING MECHANICS 2024年 77卷

作者： Rajak, Pijus Roy, Pronab Natl Inst Technol Durgapur Dept Civil Engn Durgapur 713209 India

In spite of recent advancements in reliability analysis, high-dimensional and low-failure probability problems remain challenging because many samples and function calls are required for an accurate result. Function calls lead to a sharp increase in computational cost in terms of time. For this reason, an active learning algorithm is proposed using Kriging metamodel, where an unsupervised algorithm is used to select training samples from random samples for the first and second iterations. Then, the metamodel is improved iteratively by enriching the concerned domain with samples near the limit state function and samples obtained from a space-filling design. Hence, rapid convergence with the minimum number of function calls occurs using this active learning algorithm. An efficient stopping criterion has been developed to avoid premature or late-mature terminations of the metamodel and to regulate the accuracy of the failure probability estimations. The efficacy of this algorithm is examined using relative error, number of function calls, and coefficient of efficiency in five examples which are based on high-dimensional and low-failure probability with random and interval variables.

关键词： Unsupervised clustering algorithm Kriging metamodel Sample selection strategy Space-filling design active learning algorithm Stopping criterion

来源：评论

学校读者我要写书评

暂无评论

Network Security Intrusion Detection Methods Combining Optimization algorithms and Neural Networks

引用

Procedia Computer Science 2023年 228卷 582-592页

作者： Lan Xia Xuefei Xia Jilin Communications Polytechnic Changchun 130000 China

Due to the diverse and volatile nature of today's network attacks, it is difficult to achieve satisfactory results using a single detection method. This project plans to study the network intrusion detection algorithm based on Semi-Supervised learning and active learning to improve its accuracy and effectively intercept various network attacks. First, network attack data is collected, attack characteristics are extracted, and Semi-Supervised learning method is used to cluster attack data. Secondly, by studying the clustering data, a network intrusion detection classifier based on BP neural network was constructed, and the construction process of the classifier was improved. On this basis, the effectiveness of the algorithm was verified through simulation experiments using standard datasets. The experimental results show that the combination rate of the optimized algorithm for network intrusion detection and neural network methods is over 95%, which can meet the actual accuracy requirements of network security protection. This is because the method proposed by the author combines semi supervised technology with BP neural network, which can effectively identify different types of network intrusion behavior, thereby reducing the missed detection rate and false alarm rate of network intrusion detection, and has significant advantages.

关键词： Intrusion detection BP neural network active learning algorithm Data clustering processing

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：