On the basis of machine leaning,suitable algorithms can make advanced time series analysis.This paper proposes a complex k-nearest neighbor(KNN)model for predicting financial time series.This model uses a complex feat...On the basis of machine leaning,suitable algorithms can make advanced time series analysis.This paper proposes a complex k-nearest neighbor(KNN)model for predicting financial time series.This model uses a complex feature extraction process integrating a forward rolling empirical mode decomposition(EMD)for financial time series signal analysis and principal component analysis(PCA)for the dimension reduction.The information-rich features are extracted then input to a weighted KNN classifier where the features are weighted with PCA loading.Finally,prediction is generated via regression on the selected nearest neighbors.The structure of the model as a whole is original.The test results on real historical data sets confirm the effectiveness of the models for predicting the Chinese stock index,an individual stock,and the EUR/USD exchange rate.展开更多
针对三维激光点云线性K最近邻(K-nearest neighbor, KNN)搜索耗时长的问题,提出了一种利用多处理器片上系统(multi-processor system on chip, MPSoC)现场可编程门阵列(field-programmable gate array,FPGA)实现三维激光点云KNN快速搜...针对三维激光点云线性K最近邻(K-nearest neighbor, KNN)搜索耗时长的问题,提出了一种利用多处理器片上系统(multi-processor system on chip, MPSoC)现场可编程门阵列(field-programmable gate array,FPGA)实现三维激光点云KNN快速搜索的方法。首先给出了三维激光点云KNN算法的MPSoC FPGA实现框架;然后详细阐述了每个模块的设计思路及实现过程;最后利用MZU15A开发板和天眸16线旋转机械激光雷达搭建了测试平台,完成了三维激光点云KNN算法MPSoC FPGA加速的测试验证。实验结果表明:基于MPSoC FPGA实现的三维激光点云KNN算法能在保证邻近点搜索精度的情况下,减少邻近点搜索耗时。展开更多
针对现有入侵检测技术的不足,文章研究了基于机器学习的异常入侵检测系统,将多标记和半监督学习应用于入侵检测,提出了一种基于多标记学习的入侵检测算法。该算法采用"k近邻"分类准则,统计近邻样本的类别标记信息,通过最大化...针对现有入侵检测技术的不足,文章研究了基于机器学习的异常入侵检测系统,将多标记和半监督学习应用于入侵检测,提出了一种基于多标记学习的入侵检测算法。该算法采用"k近邻"分类准则,统计近邻样本的类别标记信息,通过最大化后验概率(maximum a posteriori,MAP)的方式推理未标记数据的所属集合。在KDD CUP99数据集上的仿真结果表明,该算法能有效地改善入侵检测系统的性能。展开更多
随着互联网的快速发展,网络安全越来越受到人们的重视。传统的异常流量检测模型虽然具有较好的识别率,但需要大量有标记的数据进行训练。因此,基于无监督学习的网络异常流量检测方法被广泛采用。近年来,随着深度学习算法在异常检测中的...随着互联网的快速发展,网络安全越来越受到人们的重视。传统的异常流量检测模型虽然具有较好的识别率,但需要大量有标记的数据进行训练。因此,基于无监督学习的网络异常流量检测方法被广泛采用。近年来,随着深度学习算法在异常检测中的运用,无监督深度学习模型也不同程度地提升了检测算法的性能。然而,无监督深度学习方法往往无法避免异常检测阈值选择的问题。因此,针对现有数据标记困难和阈值选择的问题,文中提出了一种基于代价敏感度改进的K近邻算法结合阈值选择方法的异常流量检测系统。该系统不但可以准确识别恶意流量,也无需有标记数据集,极大减少了人工标注数据的工作量。实验使用UNSW-NB15、NSL-KDD和CICIDS2017数据集来验证模型的适用性,并分别与经典的机器学习算法One Class SVM以及深度学习方法AutoEncoder进行了对比。实验结果表明,在3类数据集上,与深度学习算法和传统的无监督机器学习算法相比,该算法有效提升了网络异常流量检测的性能。展开更多
Filament-induced breakdown spectroscopy(FIBS)combined with machine learning algorithms was used to identify five aluminum alloys.To study the effect of the distance between focusing lens and target surface on the iden...Filament-induced breakdown spectroscopy(FIBS)combined with machine learning algorithms was used to identify five aluminum alloys.To study the effect of the distance between focusing lens and target surface on the identification accuracy of aluminum alloys,principal component analysis(PCA)combined with support vector machine(SVM)and Knearest neighbor(KNN)was used.The intensity and intensity ratio of fifteen lines of six elements(Fe,Si,Mg,Cu,Zn,and Mn)in the FIBS spectrum were selected.The distances between the focusing lens and the target surface in the pre-filament,filament,and post-filament were 958 mm,976 mm,and 1000 mm,respectively.The source data set was fifteen spectral line intensity ratios,and the cumulative interpretation rates of PC1,PC2,and PC3 were 97.22%,98.17%,and 95.31%,respectively.The first three PCs obtained by PCA were the input variables of SVM and KNN.The identification accuracy of the different positions of focusing lens and target surface was obtained,and the identification accuracy of SVM and KNN in the filament was 100%and 90%,respectively.The source data set of the filament was obtained by PCA for the first three PCs,which were randomly selected as the training set and test set of SVM and KNN in 3:2.The identification accuracy of SVM and KNN was 97.5%and 92.5%,respectively.The research results can provide a reference for the identification of aluminum alloys by FIBS.展开更多
基金supported by the Social Science Foundation of China under Grant No.17BGL231。
文摘On the basis of machine leaning,suitable algorithms can make advanced time series analysis.This paper proposes a complex k-nearest neighbor(KNN)model for predicting financial time series.This model uses a complex feature extraction process integrating a forward rolling empirical mode decomposition(EMD)for financial time series signal analysis and principal component analysis(PCA)for the dimension reduction.The information-rich features are extracted then input to a weighted KNN classifier where the features are weighted with PCA loading.Finally,prediction is generated via regression on the selected nearest neighbors.The structure of the model as a whole is original.The test results on real historical data sets confirm the effectiveness of the models for predicting the Chinese stock index,an individual stock,and the EUR/USD exchange rate.
文摘针对现有入侵检测技术的不足,文章研究了基于机器学习的异常入侵检测系统,将多标记和半监督学习应用于入侵检测,提出了一种基于多标记学习的入侵检测算法。该算法采用"k近邻"分类准则,统计近邻样本的类别标记信息,通过最大化后验概率(maximum a posteriori,MAP)的方式推理未标记数据的所属集合。在KDD CUP99数据集上的仿真结果表明,该算法能有效地改善入侵检测系统的性能。
文摘随着互联网的快速发展,网络安全越来越受到人们的重视。传统的异常流量检测模型虽然具有较好的识别率,但需要大量有标记的数据进行训练。因此,基于无监督学习的网络异常流量检测方法被广泛采用。近年来,随着深度学习算法在异常检测中的运用,无监督深度学习模型也不同程度地提升了检测算法的性能。然而,无监督深度学习方法往往无法避免异常检测阈值选择的问题。因此,针对现有数据标记困难和阈值选择的问题,文中提出了一种基于代价敏感度改进的K近邻算法结合阈值选择方法的异常流量检测系统。该系统不但可以准确识别恶意流量,也无需有标记数据集,极大减少了人工标注数据的工作量。实验使用UNSW-NB15、NSL-KDD和CICIDS2017数据集来验证模型的适用性,并分别与经典的机器学习算法One Class SVM以及深度学习方法AutoEncoder进行了对比。实验结果表明,在3类数据集上,与深度学习算法和传统的无监督机器学习算法相比,该算法有效提升了网络异常流量检测的性能。
基金Project supported by the Natural Science Foundation of Jilin Province,China(Grant No.2020122348JC)。
文摘Filament-induced breakdown spectroscopy(FIBS)combined with machine learning algorithms was used to identify five aluminum alloys.To study the effect of the distance between focusing lens and target surface on the identification accuracy of aluminum alloys,principal component analysis(PCA)combined with support vector machine(SVM)and Knearest neighbor(KNN)was used.The intensity and intensity ratio of fifteen lines of six elements(Fe,Si,Mg,Cu,Zn,and Mn)in the FIBS spectrum were selected.The distances between the focusing lens and the target surface in the pre-filament,filament,and post-filament were 958 mm,976 mm,and 1000 mm,respectively.The source data set was fifteen spectral line intensity ratios,and the cumulative interpretation rates of PC1,PC2,and PC3 were 97.22%,98.17%,and 95.31%,respectively.The first three PCs obtained by PCA were the input variables of SVM and KNN.The identification accuracy of the different positions of focusing lens and target surface was obtained,and the identification accuracy of SVM and KNN in the filament was 100%and 90%,respectively.The source data set of the filament was obtained by PCA for the first three PCs,which were randomly selected as the training set and test set of SVM and KNN in 3:2.The identification accuracy of SVM and KNN was 97.5%and 92.5%,respectively.The research results can provide a reference for the identification of aluminum alloys by FIBS.
