Geometry of the Restricted Boltzmann Machine

被引:0
|
作者
Cueto, Maria Angelica [1 ]
Morton, Jason [2 ]
Sturmfels, Bernd [1 ]
机构
[1] Univ Calif Berkeley, Dept Math, Berkeley, CA 94720 USA
[2] Penn State Univ, Dept Math & Stat, University Pk, PA 16802 USA
关键词
Algebraic statistics; tropical geometry; deep belief network; Hadamard product; secant variety; Segre variety; inference function; linear threshold function; STATISTICAL-MODELS; SECANT VARIETIES; NETWORKS;
D O I
暂无
中图分类号
O1 [数学];
学科分类号
0701 ; 070101 ;
摘要
The restricted Boltzmann machine is a graphical model for binary random variables. Based on a complete bipartite graph separating hidden and observed variables, it is the binary analog to the factor analysis model. We study this graphical model from the perspectives of algebraic statistics and tropical geometry, starting with the observation that its Zariski closure is a Hadamard power of the first secant variety of the Segre variety of projective lines. We derive a dimension formula for the tropicalized model, and we use it to show that the restricted Boltzmann machine is identifiable in many cases. Our methods include coding theory and geometry of linear threshold functions.
引用
收藏
页码:135 / +
页数:4
相关论文
共 50 条
  • [1] Mining Invariance in Restricted Boltzmann Machine via Information Geometry
    Wang, Hailin
    Wu, Jingyi
    Ruan, Xingmao
    Hou, Yuexian
    Zhang, Peng
    2013 SIXTH INTERNATIONAL CONFERENCE ON ADVANCED COMPUTATIONAL INTELLIGENCE (ICACI), 2013, : 287 - 292
  • [2] An Infinite Restricted Boltzmann Machine
    Cote, Marc-Alexandre
    Larochelle, Hugo
    NEURAL COMPUTATION, 2016, 28 (07) : 1265 - 1288
  • [3] Tensorizing Restricted Boltzmann Machine
    Ju, Fujiao
    Sun, Yanfeng
    Gao, Junbin
    Antolovich, Michael
    Dong, Junliang
    Yin, Baocai
    ACM TRANSACTIONS ON KNOWLEDGE DISCOVERY FROM DATA, 2019, 13 (03)
  • [4] Universal Approximation Results for the Temporal Restricted Boltzmann Machine and the Recurrent Temporal Restricted Boltzmann Machine
    Odense, Simon
    Edwards, Roderick
    JOURNAL OF MACHINE LEARNING RESEARCH, 2016, 17 : 1 - 21
  • [5] Universal approximation results for the temporal restricted boltzmann machine and the recurrent temporal restricted Boltzmann machine
    Odense, Simon
    Edwards, Roderick
    Journal of Machine Learning Research, 2016, 17
  • [6] RESTRICTED BOLTZMANN MACHINE IMAGE COMPRESSION
    Kuechhold, Markus
    Simon, Maik
    Sikora, Thomas
    2018 PICTURE CODING SYMPOSIUM (PCS 2018), 2018, : 243 - 247
  • [7] Matrix Variate Restricted Boltzmann Machine
    Qi, Guanglei
    Sun, Yanfeng
    Gao, Junbin
    Hu, Yongli
    Li, Jinghua
    2016 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2016, : 389 - 395
  • [8] Decreasing the Size of the Restricted Boltzmann Machine
    Saito, Yohei
    Kato, Takuya
    NEURAL COMPUTATION, 2019, 31 (04) : 784 - 805
  • [9] VLSI Architectures for the Restricted Boltzmann Machine
    Yuan, Bo
    Parhi, Keshab K.
    ACM JOURNAL ON EMERGING TECHNOLOGIES IN COMPUTING SYSTEMS, 2017, 13 (03)
  • [10] A Removing Redundancy Restricted Boltzmann Machine
    Jiang, Yun
    Xiao, Jize
    Liu, Xi
    Hou, Jinquan
    PROCEEDINGS OF 2018 TENTH INTERNATIONAL CONFERENCE ON ADVANCED COMPUTATIONAL INTELLIGENCE (ICACI), 2018, : 57 - 62