Learning Laplacian Matrix in Smooth Graph Signal Representations

被引:412
|
作者
Dong, Xiaowen [1 ]
Thanou, Dorina [2 ]
Frossard, Pascal [2 ]
Vandergheynst, Pierre [2 ]
机构
[1] MIT, Media Lab, Cambridge, MA 02139 USA
[2] Ecole Polytech Fed Lausanne, Signal Proc Labs, LTS4, LTS2, CH-1015 Lausanne, Switzerland
基金
瑞士国家科学基金会;
关键词
Laplacian matrix learning; graph signal processing; representation theory; factor analysis; Gaussian prior; VARIABLE SELECTION; BRAIN CONNECTIVITY; MODEL; WAVELET;
D O I
10.1109/TSP.2016.2602809
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
The construction of a meaningful graph plays a crucial role in the success of many graph-based representations and algorithms for handling structured data, especially in the emerging field of graph signal processing. However, a meaningful graph is not always readily available from the data, nor easy to define depending on the application domain. In particular, it is often desirable in graph signal processing applications that a graph is chosen such that the data admit certain regularity or smoothness on the graph. In this paper, we address the problem of learning graph Laplacians, which is equivalent to learning graph topologies, such that the input data form graph signals with smooth variations on the resulting topology. To this end, we adopt a factor analysis model for the graph signals and impose a Gaussian probabilistic prior on the latent variables that control these signals. We show that the Gaussian prior leads to an efficient representation that favors the smoothness property of the graph signals. We then propose an algorithm for learning graphs that enforces such property and is based on minimizing the variations of the signals on the learned graph. Experiments on both synthetic and real world data demonstrate that the proposed graph learning framework can efficiently infer meaningful graph topologies from signal observations under the smoothness prior.
引用
收藏
页码:6160 / 6173
页数:14
相关论文
共 50 条
  • [21] DYNAMIC GRAPH LEARNING BASED ON GRAPH LAPLACIAN
    Jiang, Bo
    Yu, Yiyi
    Krim, Hamid
    Smith, Spencer L.
    2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021), 2021, : 1090 - 1094
  • [22] Laplacian Autoencoders for Learning Stochastic Representations
    Miani, Marco
    Warburg, Frederik
    Moreno-Munoz, Pablo
    Detlefsen, Nicke Skafte
    Hauberg, Soren
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35 (NEURIPS 2022), 2022,
  • [23] Graph domain adaptation with localized graph signal representations
    Pilavci, Yusuf Yigit
    Guneyi, Eylem Tugce
    Cengiz, Cemil
    Vural, Elif
    PATTERN RECOGNITION, 2024, 155
  • [24] On the spectrum of the net Laplacian matrix of a signed graph
    Stanic, Zoran
    BULLETIN MATHEMATIQUE DE LA SOCIETE DES SCIENCES MATHEMATIQUES DE ROUMANIE, 2020, 63 (02): : 205 - 213
  • [25] Matrix tree theorem for the net Laplacian matrix of a signed graph
    Mallik, Sudipta
    LINEAR & MULTILINEAR ALGEBRA, 2024, 72 (07): : 1138 - 1152
  • [26] On the spectral radius of the adjacency matrix and signless Laplacian matrix of a graph
    Jahanbani, A.
    Sheikholeslami, S. M.
    LINEAR & MULTILINEAR ALGEBRA, 2022, 70 (21): : 6846 - 6851
  • [27] On the spectrum of the quasi-Laplacian matrix of a graph
    Tan Xuezhong
    Liu Bolian
    AUSTRALASIAN JOURNAL OF COMBINATORICS, 2006, 34 : 49 - 55
  • [28] Approximations of the Generalized Inverse of the Graph Laplacian Matrix
    Bozzo, Enrico
    Franceschet, Massimo
    INTERNET MATHEMATICS, 2012, 8 (04) : 456 - 481
  • [29] On the spread of the distance signless Laplacian matrix of a graph
    Pirzada, S.
    Haq, Mohd Abrar Ul
    ACTA UNIVERSITATIS SAPIENTIAE INFORMATICA, 2023, 15 (01) : 38 - 45
  • [30] A note on unimodular congruence of the Laplacian matrix of a graph
    Liang, Hao
    Pan, Yong-Liang
    Wang, Jian
    Xu, Jun-Ming
    LINEAR & MULTILINEAR ALGEBRA, 2010, 58 (04): : 497 - 501