Accelerated stochastic multiplicative update with gradient averaging for nonnegative matrix factorizations
Accelerated stochastic multiplicative update with gradient averaging for nonnegative matrix factorizations
复制标题
DOI：
10.23919/eusipco.2018.8553610
复制
发表时间：
2018-09
期刊：
影响因子：
--
通讯作者：
Hiroyuki Kasai
中科院分区：
文献类型：
--
作者：
Hiroyuki Kasai
非负矩阵分解 (NMF) 是数据分析中的强大工具，它通过从高维数据中发现潜在特征和基于部分的模式，并且是因子矩阵具有低秩非负约束的特殊情况。将 NMF 应用于大尺寸矩阵，我们专门解决随机乘法更新（MU）规则，这是最流行的，但具有缓慢的收敛特性。本文在随机MU规则上引入了随机梯度的梯度平均技术，并提出了一种加速随机乘法更新规则：SAGMU。使用合成数据集和真实数据集进行的大量计算实验证明了 SAGMU 的有效性。
Nonnegative matrix factorization (NMF) is a powerful tool in data analysis by discovering latent features and part-based patterns from high-dimensional data, and is a special case in which factor matrices have low-rank nonnegative constraints. Applying NMF into huge-size matrices, we specifically address stochastic multiplicative update (MU) rule, which is the most popular, but which has slow convergence property. This present paper introduces a gradient averaging technique of stochastic gradient on the stochastic MU rule, and proposes an accelerated stochastic multiplicative update rule: SAGMU. Extensive computational experiments using both synthetic and real-world datasets demonstrate the effectiveness of SAGMU.