In this paper, the exact analytical solution of the rectangular plate having simplysupported segments mixed with free segments of straight edges are first given by means of the method of reciprocal theorem.By comparis...In this paper, the exact analytical solution of the rectangular plate having simplysupported segments mixed with free segments of straight edges are first given by means of the method of reciprocal theorem.By comparison .we calculate the same question by finite element method.Thecomparison shows that the analytical solution is correct.展开更多
Image segmentation is a necessary step in image analysis. Support vector machine (SVM) approach is proposed to segment images and its segmentation performance is evaluated. Experimental results show that: the effec...Image segmentation is a necessary step in image analysis. Support vector machine (SVM) approach is proposed to segment images and its segmentation performance is evaluated. Experimental results show that: the effects of kernel function and model parameters on the segmentation performance are significant; SVM approach is less sensitive to noise in image segmentation; The segmentation performance of SVM approach is better than that of back-propagation multi-layer perceptron (BP-MLP) approach and fuzzy c-means (FCM) approach.展开更多
在机载锂电池失效识别等样本不平衡的应用场景中,支持向量机(support vector machine,SVM)算法存在分离超平面偏移的问题,为此,提出分段惩罚参数支持向量机(segmented penalty parameters support vector machine,SPP-SVM)算法.该算法...在机载锂电池失效识别等样本不平衡的应用场景中,支持向量机(support vector machine,SVM)算法存在分离超平面偏移的问题,为此,提出分段惩罚参数支持向量机(segmented penalty parameters support vector machine,SPP-SVM)算法.该算法在训练过程中对样本进行分段,并根据各段内样本的识别误差自动调整惩罚参数,从而抑制超平面偏移;基于容量增量分析和灰色关联分析等方法提取并筛选特征,进而基于SPP-SVM算法建立锂电池失效识别模型;以NASA锂电池数据集和加州大学欧文分校(University of California Irvine,UCI)数据集为对象,开展对比实验.研究结果表明:与结合寻优算法的SVM相比,SPP-SVM算法识别性能更好,在不平衡程度较大的锂电池数据上,查准率和查全率的调和平均数(F1值)提升11.7%;在锂电池数据集和UCI数据集上的训练耗时缩短,减少幅度超过10倍;证明在样本不平衡情况下,使用SPP-SVM算法能够有效抑制分离超平面偏移,提升识别效果.展开更多
为了解决直推式支持向量机(transductive support vector machines,TSVM)在样本选择自动化程度低和特征学习充分性不足的问题,提出了一种融合深度空间特征与传统影像对象特征的TSVM自动高分遥感影像变化检测方法。首先,采用基于分形网...为了解决直推式支持向量机(transductive support vector machines,TSVM)在样本选择自动化程度低和特征学习充分性不足的问题,提出了一种融合深度空间特征与传统影像对象特征的TSVM自动高分遥感影像变化检测方法。首先,采用基于分形网络演化算法的叠置分割获取多时相高分遥感影像的影像对象,通过卷积神经网络提取遥感影像的深度空间特征,并与灰度、指数和纹理等传统影像对象特征联合构建特征空间;然后,利用卡方变换计算多维特征的加权特征差异度,采用最大期望算法和贝叶斯最小错误判别规则得到二值分割结果,依据变化概率自动将分割结果中准确率较高的部分标记为训练样本;最后,采用标记训练样本获得TSVM的多维特征空间二值分割超平面,进而完成自动变化检测。选择武汉市的两组高分数据集作为实验数据。实验结果表明,该方法能够实现样本自动选择,并且通过融合深度空间特征可以有效提高特征学习的充分性,平均准确率达到了88.84%,平均漏检率较仅利用传统影像对象特征的TSVM法降低了3.29个百分点,在定性和定量的变化检测有效性评价中均得到了提高。展开更多
文摘In this paper, the exact analytical solution of the rectangular plate having simplysupported segments mixed with free segments of straight edges are first given by means of the method of reciprocal theorem.By comparison .we calculate the same question by finite element method.Thecomparison shows that the analytical solution is correct.
基金Supported by the National Natural Science Foundation of China (No. 60475024)
文摘Image segmentation is a necessary step in image analysis. Support vector machine (SVM) approach is proposed to segment images and its segmentation performance is evaluated. Experimental results show that: the effects of kernel function and model parameters on the segmentation performance are significant; SVM approach is less sensitive to noise in image segmentation; The segmentation performance of SVM approach is better than that of back-propagation multi-layer perceptron (BP-MLP) approach and fuzzy c-means (FCM) approach.
文摘在机载锂电池失效识别等样本不平衡的应用场景中,支持向量机(support vector machine,SVM)算法存在分离超平面偏移的问题,为此,提出分段惩罚参数支持向量机(segmented penalty parameters support vector machine,SPP-SVM)算法.该算法在训练过程中对样本进行分段,并根据各段内样本的识别误差自动调整惩罚参数,从而抑制超平面偏移;基于容量增量分析和灰色关联分析等方法提取并筛选特征,进而基于SPP-SVM算法建立锂电池失效识别模型;以NASA锂电池数据集和加州大学欧文分校(University of California Irvine,UCI)数据集为对象,开展对比实验.研究结果表明:与结合寻优算法的SVM相比,SPP-SVM算法识别性能更好,在不平衡程度较大的锂电池数据上,查准率和查全率的调和平均数(F1值)提升11.7%;在锂电池数据集和UCI数据集上的训练耗时缩短,减少幅度超过10倍;证明在样本不平衡情况下,使用SPP-SVM算法能够有效抑制分离超平面偏移,提升识别效果.
文摘为了解决直推式支持向量机(transductive support vector machines,TSVM)在样本选择自动化程度低和特征学习充分性不足的问题,提出了一种融合深度空间特征与传统影像对象特征的TSVM自动高分遥感影像变化检测方法。首先,采用基于分形网络演化算法的叠置分割获取多时相高分遥感影像的影像对象,通过卷积神经网络提取遥感影像的深度空间特征,并与灰度、指数和纹理等传统影像对象特征联合构建特征空间;然后,利用卡方变换计算多维特征的加权特征差异度,采用最大期望算法和贝叶斯最小错误判别规则得到二值分割结果,依据变化概率自动将分割结果中准确率较高的部分标记为训练样本;最后,采用标记训练样本获得TSVM的多维特征空间二值分割超平面,进而完成自动变化检测。选择武汉市的两组高分数据集作为实验数据。实验结果表明,该方法能够实现样本自动选择,并且通过融合深度空间特征可以有效提高特征学习的充分性,平均准确率达到了88.84%,平均漏检率较仅利用传统影像对象特征的TSVM法降低了3.29个百分点,在定性和定量的变化检测有效性评价中均得到了提高。