最近中心PRQ取样的Nystr?m核递归最小二乘
作者:
作者单位:

西南大学 电子信息工程学院

中图分类号:

TN911.7

基金项目:

国家自然科学(62071391);重庆市自然科学(cstc2020jcyj-msxmX0234);中央高校基本科研业务费(2020jd001)。


Kernel recursive least squares with the nearest PRQ Nystr?m method
Author:
Affiliation:

College of Electronic Information Engineering,Southwest University

Fund Project:

The National Natural Science Foundation of China (62071391); The National Science Foundation of Chongqing (cstc2020jcyj-msxmX0234); The Central Basic Scientific Research in Colleges and Universities Operating Expenses(2020jd001).

  • 摘要
  • | |
  • 访问统计
  • |
  • 参考文献 [24]
  • | |
  • 引证文献
  • | |
  • 文章评论
    摘要:

    核递归最小二乘(KRLS, kernel recursive least squares)算法以其快速的收敛速度和较高的滤波精度等优点而闻名于核自适应滤波器(KAFs, kernel adaptive filters)的算法中,然而其发展却受到线性增长结构和巨大计算复杂度的限制。为了缓解这些问题,基于概率密度秩量化取样的Nystr?m的核递归最小二乘方法(NysKRLS-PRQ, kernel recursive least squares with the Nystr?m method based on probability density rank-based quantization sampling)被提出,该算法以牺牲一定的滤波精度为代价来降低KRLS的计算复杂度。为进一步提高NysKRLS-PRQ的滤波精度,提出了基于最近中心PRQ取样Nystr?m方法的核递归最小二乘(NNKRLS-PRQ, kernel recursive least squares with nearest PRQ Nystr?m method)算法。在NNKRLS-PRQ中,特征空间中的输入向量被分为 个不相交的簇。根据输入向量的分布情况以及高斯核函数的置信概率,NNKRLS-PRQ能够自适应地从对应的簇中选择合适的样本来映射输入向量。通过仿真实验证明了NNKRLS-PRQ算法的优越性。

    Abstract:

    The development of the kernel recursive least squares (KRLS) algorithm, popular with the fastest convergence rate and best filtering accuracy in kernel adaptive filters (KAFs), is limited by the linear growing structure and enormous computational complexity. To alleviate these issues, KRLS with the Nystr?m method (NysKRLS) based on probability density rank-based quantization sampling (NysKRLS-PRQ) has been proposed to alleviate the complexity of KRLS at the cost of decreasing accuracy. To further improve the accuracy of NysKRLS-PRQ, we propose KRLS with nearest PRQ Nystr?m method (NNKRLS-PRQ) algorithm was proposed. The input vectors of NNKRLS-PRQ in the feature space are organized by p disjoint clusters. According to the distribution situation of input vectors and the confidence probability of the Gaussian kernel, NNKRLS-PRQ adaptively chooses proper samples selected from the corresponding cluster to project the input vector. The simulations prove the superiorities of NKMNKRLS.

    参考文献
    [1] Liu W F, Principe J C, Haykin S. Kernel adaptive filtering[M]. New Jersey: John Wiley Sons Ltd , 2010: 95-96.
    [2] Mitchell T M. Machine learning[M]. New York, NY, USA: McGraw-Hill, 1997.
    [3] Liu Y Q, Chao S, Jiang S D. A reduced gaussian kernel least-mean-square algorithm for nonlinear adaptive signal processing[J]. Circuits, Systems, and Signal Process, 2019, 38(1): 371-394.
    [4] Scarpiniti M, Comminiello D, Parisi R, et al. Nonlinear spline adaptive filtering[J]. Signal Processing, 2013, 93(4): 772-783.
    [5] Meher P K, Park S Y. Critical-path analysis and low-complexity implementation of the LMS adaptive algorithm[J]. IEEE Transactions on Circuits and Systems I: Regular Paper, 2014, 61(3): 778-788.
    [6] 赵知劲, 张笑菲. 核递归最小P范数算法[J]. 信号处理, 2017, 33(4): 523-527.Zhao Z J, Zhang X F. Kernel recursive least mean P-norm algorithm[J]. Journal of Signal Processing, 2017, 33(4): 523-527. (in Chinese)
    [7] Shin H C, Sayed A H. Mean-square performance of a family of affine projection algorithms[J]. IEEE Transactions on Signal Processing, 2004, 52(1): 90-102.
    [8] Engel Y, Mannor S, Meir R. The kernel recursive least-squares algorithm[J]. IEEE Transactions on Signal Processing, 2004, 52(8): 2275-2285.
    [9] Platt J. A resource-allocating network for function interpolation[J]. Neural Computation, 1991, 3(2): 213-225.
    [10] 欧阳城添, 袁瑾. 基于学习矢量量化的空调压缩机声纹诊断方法[J]. 计算机与工程设计, 2021, 42(09): 2634-2641.Ouyang C T, Yuan J. Voiceprint diagnosis method of air conditioning compressor based on learning vector quantization[J]. Computer Engineering and Design, 2021, 42(09): 2634-2641. (in Chinese)
    [11] 汪可, 杨丽君, 廖瑞金, 等. 动态时间规整算法在局部放电模式识别中的应用[J]. 重庆大学学报(自然科学版), 2011, 34(12): 54-60.Wang K, Yang L J, Liao R J, et al. Application of dynamic time warping algorithm to partial discharge pattern recognition[J]. Journal of Chongqing University (Natural Science Edition), 2011, 34(12): 54-60. (in Chinese)
    [12] Chen T, Wang S Y. An efficient nonlinear dichotomous coordinate descent adaptive algorithm based on random Fourier features[J]. IEEE Signal Processing Letters, 2020, 27: 1804- 1808.
    [13] Zhang H N, Yang B, Wang L, et al. General cauchy conjugate gradient algorithms based on multiple random Fourier features[J]. IEEE Transactions on Signal Processing, 2021, 69: 1859-1873.
    [14] Yang T B, Li Y F, Mahdavi M, et al. Nystr?m method vs random Fourier features: a theoretical and empirical comparison[J]. Advances in Neural Information Processing Systems(NIPS), 2012, 1:476-484.
    [15] 徐杨杨, 孙建国, 商耀达. 一种利用Nystr?m离散与FFT快速褶积的散射地震波并行计算方法[J]. 地球物理学报, 2021, 64(8): 2877-2887.Xu Y Y, Sun J G, Shang Y D. A parallel computation method for scattered seismic waves using Nystr?m discretization and FFT fast convolution[J]. Chinese J. Geophys, 2021, 64(8): 2877-2887. (in Chinese)
    [16] Wang S Y, Dang L J, Qian G B, et al. Kernel recursive maximum correntropy with Nystr?m approximation[J], Neurocomputing, 2019, 329: 424-432.
    [17] Kumar S, Mohri M, Talwaikar A. Sampling methods for the Nystr?m method[J]. Journal of Machine Learning Research, 2012, 13(1): 981-1006.
    [18] Zhang T, Wang S Y, Huang X W, et al. Kernel recursive least squares algorithm based on the Nystr?m method with k-means sampling[J]. IEEE Signal Processing Letters, 2020, 27: 361-365.
    [19] 李珍萍, 赵雨薇, 张煜炜. 共同配送选址-路径优化模型与算法[J]. 重庆大学学报(自然科学版), 2020, 43(1): 28-43.Li Z P, Zhao Y W, Zhang Y W. Optimization model and algorithm of location-routing for joint distribution[J]. Journal of Chongqing University (Natural Science Edition), 2020, 43(1): 28-43. (in Chinese)
    [20] Qin Z D, Chen B D, Gu Y T, N. Zheng, et al. Probability density rank-based quantization for convex universal learning machines[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(8): 3100-3113.
    [21] Zhang T, Wang S Y. Nystr?m kernel algorithm under generalized maximum correntropy criterion[J]. IEEE Signal Processing Letters, 2020, 27: 1535-1539.
    [22] Li K, Principe J C. Transfer learning in adaptive filters: the nearest instance centroid-estimation kernel least-mean-square algorithm[J]. IEEE Transactions on Signal Processing, 2017, 65(24): 6520-6535.
    [23] Zhang H N, Wang L, Zhang T, et al. The nearest-instance-centroid-estimation kernel recursive least squares algorithms[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2020, 67(7): 1344-1348.
    [24] Zhang H N, Jiang H, Wang S Y. Kernel least mean square based on the sparse Nystr?m method[C]// 2020 IEEE International Symposium on Circuits and Systems (ISCAS). October 12-14, 2020, Seville, Spain: IEEE, 2020: 1-5.
    相似文献
    引证文献
    引证文献 [0] 您输入的地址无效!
    没有找到您想要的资源,您输入的路径无效!

    网友评论
    网友评论
    分享到微博
    发 布
引用本文
分享
文章指标
  • 点击次数:178
  • 下载次数: 0
  • HTML阅读次数: 0
  • 引用次数: 0
历史
  • 收稿日期:2021-10-16
  • 最后修改日期:2022-07-24
  • 录用日期:2022-09-08
文章二维码