Neural nonnegative matrix factorization for hierarchical multilayer topic modeling

被引：0

作者：

Haddock, Jamie ^{[1
]}

Will, Tyler ^{[2
]}

Vendrow, Joshua ^{[3
]}

Zhang, Runyu ^{[4
]}

Molitor, Denali ^{[5
]}

Needell, Deanna ^{[6
]}

Gao, Mengdi ^{[7
]}

Sadovnik, Eli ^{[1
,2
,3
,4
,5
,6
,7
]}

机构：

[1] Harvey Mudd Coll, Dept Math, 301 Platt Blvd, Claremont, CA 91711 USA

[2] Optimal Dynam, New York, NY 10001 USA

[3] MIT, Dept EECS, 50 Vassar St, Cambridge, MA 02140 USA

[4] Harvard Univ, Sch Engn & Appl Sci, 150 Western Ave, Cambridge, MA 02138 USA

[5] Google, Seattle, WA 98103 USA

[6] Univ Calif Los Angeles, Dept Math, 520 Portola Plaza, Los Angeles, CA 90095 USA

[7] Schlumberger, Menlo Pk, CA 94025 USA

来源：

SAMPLING THEORY SIGNAL PROCESSING AND DATA ANALYSIS | 2024年 / 22卷 / 01期

关键词：

Hierarchical topic models; Nonnegative matrix factorization; Backpropagation; ALGORITHMS;

D O I：

10.1007/s43670-023-00077-3

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

We introduce a new method based on nonnegative matrix factorization, Neural NMF, for detecting latent hierarchical structure in data. Datasets with hierarchical structure arise in a wide variety of fields, such as document classification, image processing, and bioinformatics. Neural NMF recursively applies NMF in layers to discover overarching topics encompassing the lower-level features. We derive a backpropagation optimization scheme that allows us to frame hierarchical NMF as a neural network. We test Neural NMF on a synthetic hierarchical dataset, the 20 Newsgroups dataset, and the MyLymeData symptoms dataset. Numerical results demonstrate that Neural NMF outperforms other hierarchical NMF methods on these data sets and offers better learned hierarchical structure and interpretability of topics.

引用

页数：38

共 50 条

[21] Sparse modeling of EELS and EDX spectral imaging data by nonnegative matrix factorization
Shiga, Motoki
Tatsumi, Kazuyoshi
Muto, Shunsuke
Tsuda, Koji
Yamamoto, Yuta
Mori, Toshiyuki
Tanji, Takayoshi
ULTRAMICROSCOPY, 2016, 170 : 43 - 59
[22] Topic Splitting: A Hierarchical Topic Model Based on Non-Negative Matrix Factorization
Rui Liu
Xingguang Wang
Deqing Wang
Yuan Zuo
He Zhang
Xianzhu Zheng
Journal of Systems Science and Systems Engineering, 2018, 27 : 479 - 496
[23] Nonnegative matrix factorization of a correlation matrix
Sonneveld, P.
van Kan, J. J. I. M.
Huang, X.
Oosterlee, C. W.
LINEAR ALGEBRA AND ITS APPLICATIONS, 2009, 431 (3-4) : 334 - 349
[24] NONNEGATIVE MATRIX FACTORIZATION WITH MATRIX EXPONENTIATION
Lyu, Siwei
2010 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, 2010, : 2038 - 2041
[25] ON THE COMPLEXITY OF NONNEGATIVE MATRIX FACTORIZATION
Vavasis, Stephen A.
SIAM JOURNAL ON OPTIMIZATION, 2009, 20 (03) : 1364 - 1377
[26] WEIGHTED NONNEGATIVE MATRIX FACTORIZATION
Kim, Yang-Deok
Choi, Seungjin
2009 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, VOLS 1- 8, PROCEEDINGS, 2009, : 1541 - 1544
[27] MINIMAX LOWER BOUNDS FOR NONNEGATIVE MATRIX FACTORIZATION
Alsan, Mine
Liu, Zhaoqiang
Tan, Vincent Y. F.
2018 IEEE STATISTICAL SIGNAL PROCESSING WORKSHOP (SSP), 2018, : 363 - 367
[28] A Framework for Compressed Weighted Nonnegative Matrix Factorization
Yahaya, Farouk
Puigt, Matthieu
Delmaire, Gilles
Roussel, Gilles
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2024, 72 : 4798 - 4811
[29] The rise of nonnegative matrix factorization: Algorithms and applications
Guo, Yi-Ting
Li, Qin-Qin
Liang, Chun-Sheng
INFORMATION SYSTEMS, 2024, 123
[30] About Nonnegative Matrix Factorization: On the posrank Approximation
de Almeida, Ana
ADAPTIVE AND NATURAL COMPUTING ALGORITHMS, PT II, 2011, 6594 : 295 - 304

← 1 2 3 4 5 →