• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Computer science and its applications >Estimating Number of Columns in Mixing Matrix for Under-Determined ICA Using Observed Signal Clustering and Exponential Filtering
【24h】

Estimating Number of Columns in Mixing Matrix for Under-Determined ICA Using Observed Signal Clustering and Exponential Filtering

机译：使用观测信号聚类和指数滤波估计欠定ICA的混合矩阵中的列数

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Under-determined Independent Component Analysis arises in a variety of signal processing applications, including speech processing. In this paper, we proposed a new approach focusing on the estimation of the number of columns and their values of the mixing matrix. The method is based on the observation that the observed vectors must be clustered along the direction of the column vectors of the mixing matrix. A new clustering measure and cluster direction finding are introduced. The propose algorithms are tested with real speech signals and compared with both AICA method and Information Index Removal, Perturbed Mean Shift Algorithm. Our result gives the correct number of columns with higher accuracy under the performance measure of algebraic matrix distance index.
机译：不确定的独立分量分析出现在包括语音处理在内的各种信号处理应用中。在本文中，我们提出了一种新方法，重点是估计混合矩阵的列数及其值。该方法基于以下观察：所观察到的向量必须沿着混合矩阵的列向量的方向聚类。介绍了一种新的聚类测度和聚类方向寻找方法。对提出的算法进行了真实语音信号的测试，并与AICA方法和信息索引去除扰动均值漂移算法进行了比较。在代数矩阵距离指数的性能指标下，我们的结果给出了具有更高准确度的正确列数。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号