共查询到19条相似文献,搜索用时 234 毫秒
1.
基本概率指派(Basic probability assignment, BPA)生成
是应用D S证据理论的关键环节和第一步,而如何生成BPA仍然是一个有待解决的问题。本文
提出一种基于云模型的BPA生成方法,首先,采用逆向云发生器生成每类样本在某属性下的
正态云模型。其次,利用前件云发生器得到待测样本在该属性下对每类样本的确定度期望。
再次,给出一种正态云模型交叠度计算方法,用确定度最大类的正态云模型与其他种类的最
大交叠度作为对全集的信任度。最后,对确定度进行归一化得到待测样本的BPA。实验结果
验证了该方法的有效性,此外,在样本数据较少情况下也能有效生成BPA。 相似文献
2.
如果对抗样本的迁移性越强,则其攻击结构未知的深度神经网络模型的效果越好,所以设计对抗样本生成方法的一个关键在于提升对抗样本的迁移性.然而现有方法所生成的对抗样本,与模型的结构和参数高度耦合,从而难以对结构未知的模型进行有效攻击.类别显著性映射能够提取出样本的关键特征信息,而且在不同网络模型中有较高的相似度.基于显著性映... 相似文献
3.
4.
近年来,变电站中广泛采用机器视觉算法分析多时相巡检图像的差异变化,用于检测各类变电设备缺陷,以确保运行安全.然而,由于拍摄时刻不同,多时相图像间存在天气、光照、季节等各类干扰变化,对变电设备的缺陷检测提出了挑战.对此,提出一种基于多时相巡检图像的变电设备抗干扰缺陷检测方法.首先,利用风格迁移模型CycleGAN学习不同风格域之间的映射关系,并基于检测图生成足量存在天气、光照、季节干扰变化的干扰图;其次,基于参考图$+$检测图$+$干扰图三元组对三重孪生网络TripleNet进行协同训练,在特征层面提出空间一致性损失以抵抗各类干扰变化,用于提取三者鲁棒的多尺度差异特征;最后,搭建特征聚合网络PANet融合多尺度差异特征,输出多尺度的缺陷检测结果.在实际变电设备多时相巡检图像数据集上进行实验验证,结果表明,所提出方法相较于非孪生网络和一般孪生网络可提升2.09%和0.67%的mAP,且在原始样本与干扰样本上的检测精度更均衡,而且所提出方法可以在提升变电设备缺陷检测模型精度的同时增强模型的抗干扰能力. 相似文献
5.
6.
针对当前黑盒环境中,主流的图像对抗攻击算法在有限的目标模型访问查询次数条件下攻击准确率低的问题,提出一种基于目标特征和限定区域采样的目标攻击算法.首先根据原始图像和目标图像生成初始对抗样本;然后在Simplex-mean噪声区域中进行扰动采样,并根据对抗样本和原始图像差异度以及目标特征区域位置决定扰动大小;最后将扰动作... 相似文献
7.
8.
9.
D-S合成方法作用的对象是基本概率指派(basic probability assign,BPA),如何生成BPA是D-S理论应用中重要且有待解决的首要步骤。针对生成BPA提出一种基于核密度估计KDE(kernel density estimation)的BPA生成方法:训练数据用于构建基于最优化窗宽的核密度估计的数据属性模型;然后利用训练数据的核密度模型计算测试数据的密度—距离—分布值Tri-D(density-distance-distribution),通过嵌套式的方法分配Tri-D值获取测试数据对应的BPA;最后D-S合成BPA得到最终判断,通过分类准确率来判断BPA生成方法的有效性。实验通过在UCI数据集上与其他方法的分类准确率对比验证了提出方法的有效性。 相似文献
10.
基于证据理论,提出一种新的区间直觉模糊集决策模型.首先采用区间直觉模糊集表示属性值,将区间直觉模糊数转换为区间BPA;然后利用基于区间数的组合规则进行融合;最后将融合后的区间BPA转换为经典BPA用于决策,可直接方便地实现多属性数据的融合.该模型的优点在于:简单直观,能更有效地反映原始信息的不确定度;通用性好,可以推广到其他区间直觉模糊集的应用领域.算例结果表明了所提出模型的有效性. 相似文献
11.
Peida Xu Xiaoyan Su Sankaran Mahadevan Chenzhao Li Yong Deng 《Applied Intelligence》2014,41(3):681-693
As an important tool for knowledge representation and decision-making under uncertainty, Dempster-Shafer evidence theory (D-S theory) has been used in many fields. The application of D-S theory is critically dependent on the availability of the basic probability assignment (BPA). The determination of BPA is still an open issue. A non-parametric method to obtain BPA is proposed in this paper. This method can handle multi-attribute datasets in classification problems. Each attribute value of the dataset sample is treated as a stochastic quantity. Its non-parametric probability density function (PDF) is calculated using the training data, which can be regarded as the probability model for the corresponding attribute. The BPA function is then constructed based on the relationship between the test sample and the probability models. The missing attribute values in datasets are treated as ignorance in the framework of the evidence theory. This method does not have the assumption of any particular distribution. As a result, it can be flexibly used in many engineering applications. The obtained BPA can avoid high conflict between evidence, which is desired in data fusion. Several benchmark classification problems are used to demonstrate the proposed method and to compare against existing methods. The constructed classifier based on the proposed method compares well to the state-of-the-art algorithms. 相似文献
12.
The theory of intuitionistic fuzzy sets (IFS) is widely used for dealing with vagueness and the Dempster--Shafer (D-S) evidence theory has a widespread use in multiple criteria decision-making problems under uncertain situation. However, there are many methods to aggregate intuitionistic fuzzy numbers (IFNs), but the aggregation operator to fuse basic probability assignment (BPA) is rare. Power average (P-A) operator, as a powerful operator, is useful and important in information fusion. Motivated by the idea of P-A power, in this paper, a new operator based on the IFS and D-S evidence theory is proposed, which is named as intuitionistic fuzzy evidential power average (IFEPA) aggregation operator. First, an IFN is converted into a BPA, and the uncertainty is measured in D-S evidence theory. Second, the difference between BPAs is measured by Jousselme distance and a satisfying support function is proposed to get the support degree between each other effectively. Then the IFEPA operator is used for aggregating the original IFN and make a more reasonable decision. The proposed method is objective and reasonable because it is completely driven by data once some parameters are required. At the same time, it is novel and interesting. Finally, an application of developed models to the ‘One Belt, One road’ investment decision-making problems is presented to illustrate the effectiveness and feasibility of the proposed operator. 相似文献
13.
判断矩阵的一致性程度可以反映专家思维的一致性水平,是基于判断矩阵确定专家权重的一个重要因素。针对传统欧氏距离方法对元素间相对差异关注不够,以及判断矩阵相似度(或差异程度)的度量易受1-9标度影响的问题,研究基于判断矩阵一致性程度的专家权重确定方法,分别提出基于欧氏距离与灰色关联度的专家权重确定方法和基于相对距离的专家权重确定方法。前者通过引入灰色关联度,计算得到的是各专家判断矩阵一致性程度比较意义上的相对值;后者通过采用相对距离方法来减弱1-9标度的影响。算例分析中,基于一致性检验指标CR验证了本文2种方法得到的专家权重排序较传统欧氏距离方法得到的排序结果合理,并分析了各种方法在专家权重具体取值上的不同特点,为今后实际应用中的方法选择提供参考。 相似文献
14.
15.
16.
针对SVM进行图像分割时存在对噪声和孤立点较敏感导致分割结果不佳和抗造性能低下等问题,提出一种基于视觉注意和改进隶属度的FSVM (Modified fuzzy SVM,MFSVM)彩色图像分割方法.该方法在考虑人类视觉显著性检测机制因素的同时,对标准的模糊SVM算法进行改进,新的隶属度函数综合考虑了样本点距离类中心的远近以及样本点的疏密程度,从而有效惩罚噪声点并增强了支持向量的作用.通过彩色图像分割进行验证,结果显示与标准的SVM及基于样本疏密程度隶属度的FSVM分割方法相比,本文方法能够对复杂场景下的彩色进行有效分割,同时呈现出良好的抗噪能力. 相似文献
17.
Many relations in the real world can be described by mathematical language. Fuzzy set theory can transform human language into mathematical language and use membership degree function to describe relations between events. Dempster–Shafer evidence theory provides basic probability assignment (BPA), which can describe the occurrence rate of attributes in basic events. Based on the known membership degree function and BPA distribution, a new evaluation method is proposed in this paper to analyze decision making. Given the relations among relevant events, which are expressed by BPA distribution and membership degree function, the relations among basic events and top event can be obtained. The Dempster's combination rule and pignistic probability transformation are used to transform BPA distribution into probability distribution. The belief measure is applied to deal with these fuzzy relations. Some numerical examples are given in this paper to illustrate the proposed evaluation methodology. 相似文献
18.
19.
针对目前灰色关联理论中模型结果检验方法的不足,提出一种对相近性灰色关联模型关联序结果稳定性检验的方法.首先,以序列间对应观测点的数值差异作为构造序列间量级大小变化的最简单元,用以表征和定义序列间的相近性灰色关联度;其次,引入信息熵的基本思想,基于序列间量级大小变化的最简单元构建信息灰熵作为相近性视角下序列间关系稳定程度的度量标准;然后,通过将研究对象的行为序列平均划分为不同阶段来计算在不同阶段变化下序列间相近性灰色关联度的变动值,以动态关联数值与信息灰熵共同构造相近性灰色关联模型的稳定性检验系数,通过比较检验系数大小最终确定模型间结果的稳定程度;最后,选取几种常见的相近性灰色关联模型对2016年~2021年南京市主要大气污染物进行影响因素分析,根据计算得到的检验系数对各模型的分析结果进行验证和筛选. 相似文献