Volume-Regularized Nonnegative Tucker Decomposition with Identifiability Guarantees

Volume-Regularized Nonnegative Tucker Decomposition with Identifiability Guarantees
复制标题

具有可识别性保证的体积正则化非负 Tucker 分解

DOI:
10.1109/icassp49357.2023.10096076
复制
发表时间:
2023
期刊:
Proceedings of the IEEE International Conference on Acoustics Speech and Signal Processing
影响因子:
--
通讯作者:
Huang, Kejun
Huang, Kejun
中科院分区:
--
文献类型:
--
作者:
Sun, Yuchen;Huang, Kejun

文献摘要

参考文献

相似文献

众所周知,多维张量的 Tucker 分解不是唯一的,因为它的因子受到类似于矩阵分解模型的旋转模糊性的影响。受到最近非负矩阵分解可识别性方面成功的启发,这项工作的目标是为非负 Tucker 分解 (NTD) 取得类似的结果。我们建议添加矩阵体积正则化作为可识别性标准,并表明如果所有 Tucker 因子满足充分分散条件,NTD 确实是可识别的。然后,我们推导出一种算法来求解 NTD 的修改公式,该算法最小化近似值的广义 Kullback-Leibler 散度加上所提出的矩阵体积正则化。数值实验表明了该方法的有效性。
It is well-known that the Tucker decomposition of a multi-dimensional tensor is not unique, because its factors are subject to rotation ambiguities similar to matrix factorization models. Inspired by the recent success in the identifiability of nonnegative matrix factorization, the goal of this work is to achieve similar results for nonnegative Tucker decomposition (NTD). We propose to add a matrix volume regularization as the identifiability criterion, and show that NTD is indeed identifiable if all of the Tucker factors satisfy the sufficiently scattered condition. We then derive an algorithm to solve the modified formulation of NTD that minimizes the generalized Kullback-Leibler divergence of the approximation plus the proposed matrix volume regularization. Numerical experiments show the effectiveness of the proposed method.
DOI: --
发表时间: 2016-11
期刊: ArXiv
影响因子: --
作者:
Kejun Huang;Xiao Fu;N. Sidiropoulos
通讯作者: Kejun Huang;Xiao Fu;N. Sidiropoulos
DOI: 10.1038/44565
发表时间: 1999-10-21
期刊: NATURE
影响因子: 64.8
作者:
Lee, DD;Seung, HS
通讯作者: Seung, HS
HOQRI:可扩展 Tucker 分解的高阶 QR 迭代
DOI: 10.1109/icassp43922.2022.9746726
发表时间: 2022
期刊: ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
影响因子: --
作者:
Yuchen Sun;Kejun Huang
通讯作者: Kejun Huang
非负Tucker分解的一些理论
DOI: 10.1007/978-3-319-53547-0_15
发表时间: 2017
影响因子: 8.4
作者:
Jeremy E. Cohen;P. Comon;Nicolas Gillis
通讯作者: Nicolas Gillis