Stable and fast update rules for independent vector analysis based on auxiliary function technique
Stable and fast update rules for independent vector analysis based on auxiliary function technique
复制标题
DOI：
10.1109/aspaa.2011.6082320
复制
发表时间：
2011-11
期刊：
影响因子：
--
通讯作者：
Nobutaka Ono
中科院分区：
文献类型：
--
作者：
Nobutaka Ono
提出了一种基于辅助函数技术的独立向量分析（IVA）稳定快速的更新规则。该算法包括两个可选的更新：1）加权协方差矩阵更新和2）分层矩阵更新，其中不包括调整参数，如步长。保证了目标函数在每次更新时的单调下降。实验结果表明，推导的更新规则产生更快的收敛速度和更好的结果比自然梯度更新。
This paper presents stable and fast update rules for independent vector analysis (IVA) based on auxiliary function technique. The algorithm consists of two alternative updates: 1) weighted covariance matrix updates and 2) demixing matrix updates, which include no tuning parameters such as step size. The monotonic decrease of the objective function at each update is guaranteed. The experimental evaluation shows that the derived update rules yield faster convergence and better results than natural gradient updates.