期刊文献+
共找到7篇文章
< 1 >
每页显示 20 50 100
Nesterov动量迭代降噪对抗攻击算法NMI-FGSM&Whey
1
作者 陶永才 李子晨 +2 位作者 石育澄 石磊 卫琳 《小型微型计算机系统》 CSCD 北大核心 2021年第7期1447-1452,共6页
目前基于深度神经网络的图像分类器易受到对抗样本的攻击,而对抗样本往往是某种算法攻击原始样本所生成的特定图像.本文针对攻击算法的弱鲁棒性及低隐蔽性提出一种基于Nesterov-Momentum动量迭代以及Whey优化的NMI-FGSM&Whey攻击算... 目前基于深度神经网络的图像分类器易受到对抗样本的攻击,而对抗样本往往是某种算法攻击原始样本所生成的特定图像.本文针对攻击算法的弱鲁棒性及低隐蔽性提出一种基于Nesterov-Momentum动量迭代以及Whey优化的NMI-FGSM&Whey攻击算法.首先在一般动量迭代攻击中加入Nesterov项,其次于生成对抗样本之后对图像实行Whey优化,在保证攻击性能的情况下有效地去除多余噪声.实验部分以Inception-v3、Resnet-152和IncRes-v2三种分类模型为基础,结合单模型攻击和融合模型攻击对比了多种攻击方法.实验结果表明所提算法在白盒模式中展现出较强的攻击力,并在有效减少噪声强度的同时于黑盒情景中呈现出较好的迁移性能. 展开更多
关键词 图像分类 对抗攻击 动量迭代 噪声
下载PDF
高阶马尔科夫链张量方程的一种反迭代算法
2
作者 周艺 吕来水 喻高航 《杭州电子科技大学学报(自然科学版)》 2022年第1期89-93,共5页
针对高阶马尔科夫链极限概率分布问题,提出一种带有动量项的反迭代算法,并在一定条件下建立该算法的收敛性。最后,通过数值算例验证了所提算法的有效性和优越性。
关键词 极限概率分布向量 转移概率张量 动量算法 张量方程
下载PDF
多组分溶质运移参数反演的数值模拟 被引量:2
3
作者 马昱 李功胜 +1 位作者 王继胜 魏镇 《山东理工大学学报(自然科学版)》 CAS 2008年第1期1-6,共6页
讨论了多孔介质中以硫酸盐为主的多组分溶质运移的数学模型.鉴于模型参数的不可直接测量性,提出了一类多参数反演问题,并应用最佳摄动量迭代算法进行了不同参数取值条件下的数值模拟.数值反演结果表明,给出的最佳摄动量迭代算法至少对... 讨论了多孔介质中以硫酸盐为主的多组分溶质运移的数学模型.鉴于模型参数的不可直接测量性,提出了一类多参数反演问题,并应用最佳摄动量迭代算法进行了不同参数取值条件下的数值模拟.数值反演结果表明,给出的最佳摄动量迭代算法至少对于所提出的多方程、多参数的反演问题是有效的. 展开更多
关键词 多组分溶质运移 对流弥散模型 水化学分析 多参数反演 最佳摄动量迭代算法 数值模拟
下载PDF
一个扰动土柱试验模型及参数反演的数值模拟 被引量:2
4
作者 魏镇 李功胜 池光胜 《山东理工大学学报(自然科学版)》 CAS 2009年第2期1-6,共6页
采用细砂(粒径约2mm)均匀填充土柱,用配置的与煤系矿坑水成份相似的淋滤液进行扰动土柱试验.通过对试验过程及穿透数据的分析,建立了一个描述淋滤液中各离子迁移行为的数学模型,并应用一种最佳摄动量迭代算法对未知的模型参数进行了数... 采用细砂(粒径约2mm)均匀填充土柱,用配置的与煤系矿坑水成份相似的淋滤液进行扰动土柱试验.通过对试验过程及穿透数据的分析,建立了一个描述淋滤液中各离子迁移行为的数学模型,并应用一种最佳摄动量迭代算法对未知的模型参数进行了数值反演模拟.结果表明,所用算法对于文中扰动土柱试验模型的参数反演是有效的. 展开更多
关键词 扰动土柱试验 数学模型 参数反演 最佳摄动量迭代算法 数值模拟
下载PDF
An Improved Control Vector Iteration Approach for Nonlinear Dynamic Optimization. II. Problems with Path Constraints 被引量:1
5
作者 胡云卿 刘兴高 薛安克 《Chinese Journal of Chemical Engineering》 SCIE EI CAS CSCD 2014年第2期141-145,共5页
This paper considers dealing with path constraints in the framework of the improved control vector iteration (CVI) approach. Two available ways for enforcing equality path constraints are presented, which can be dir... This paper considers dealing with path constraints in the framework of the improved control vector iteration (CVI) approach. Two available ways for enforcing equality path constraints are presented, which can be directly incorporated into the improved CVI approach. Inequality path constraints are much more difficult to deal with, even for small scale problems, because the time intervals where the inequality path constraints are active are unknown in advance. To overcome the challenge, the ll penalty function and a novel smoothing technique are in-troduced, leading to a new effective approach. Moreover, on the basis of the relevant theorems, a numerical algo-rithm is proposed for nonlinear dynamic optimization problems with inequality path constraints. Results obtained from the classic batch reaCtor operation problem are in agreement with the literature reoorts, and the comoutational efficiency is also high. 展开更多
关键词 nonlinear dynamic optimization control vector iteration path constraint penalty function method
下载PDF
Nonlinear vibration of an elastically restrained tapered beam 被引量:2
6
作者 KARIMPOUR S. GANJI S.S. +2 位作者 BARARI A. IBSEN L.B. DOMAIRRY G. 《Science China(Physics,Mechanics & Astronomy)》 SCIE EI CAS 2012年第10期1925-1930,共6页
This paper presents the analytical simulation of an elastically restrained tapered cantilever beam using the energy balance method(EBM) and the iteration perturbation method(IPM).To assess the accuracy of solutions,we... This paper presents the analytical simulation of an elastically restrained tapered cantilever beam using the energy balance method(EBM) and the iteration perturbation method(IPM).To assess the accuracy of solutions,we compare the results with the harmonic balance method(HBM).The obtained results from EBM and IPM are in excellent agreement with HBM results.The results show that both methods can be easily extended to other nonlinear oscillations and it can be predicted that both methods can be found widely applicable in engineering and physics. 展开更多
关键词 energy balance method (EBM) iteration perturbation method (IPM) nonlinear oscillation
原文传递
Research on the strategy of angles-only relative navigation for autonomous rendezvous 被引量:13
7
作者 LI JiuRen LI HaiYang +1 位作者 TANG GuoJin LUO YaZhong 《Science China(Technological Sciences)》 SCIE EI CAS 2011年第7期1865-1872,共8页
This paper considers the problem of angles-only relative navigation for autonomous rendezvous. Methods for determining degree of observability (DO0) and latent range information of orbital maneuver are proposed for ... This paper considers the problem of angles-only relative navigation for autonomous rendezvous. Methods for determining degree of observability (DO0) and latent range information of orbital maneuver are proposed for analyzing and enhancing the precision of relative position and velocity estimation. The equations of angles-only relative navigation are set forth on the con- dition that optical camera is the only viable sensor for relative measurement, and expressions for the DO0 of relative navigation are obtained by using the Newton iterative method. The latent range information of orbital maneuver is analyzed, which is employed to enhance the DOO of angles-only relative navigation. Simulation result shows that DOO is effective to describe the observability level of relative position and velocity, and the latent range information is useful in enhancing the DOO of the angles-only relative navigation. 展开更多
关键词 autonomous rendezvous relative navigation angles-only degree of observability orbital maneuver
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部