Bayesian Inference over the Stiefel Manifold via the Givens Representation

被引:4
|
作者
Pourzanjani, Arya A. [1 ]
Jiang, Richard M. [1 ]
Mitchell, Brian [1 ]
Atzberger, Paul J. [2 ]
Petzold, Linda R. [1 ]
机构
[1] Univ Calif Santa Barbara, Comp Sci Dept, Santa Barbara, CA 93106 USA
[2] Univ Calif Santa Barbara, Math Dept, Santa Barbara, CA 93106 USA
来源
BAYESIAN ANALYSIS | 2021年 / 16卷 / 02期
基金
美国国家科学基金会;
关键词
orthogonal matrix; transformation; dimensionality reduction; principal component analysis; SIMULATION;
D O I
10.1214/20-BA1202
中图分类号
O1 [数学];
学科分类号
0701 ; 070101 ;
摘要
We introduce an approach based on the Givens representation for posterior inference in statistical models with orthogonal matrix parameters, such as factor models and probabilistic principal component analysis (PPCA). We show how the Givens representation can be used to develop practical methods for transforming densities over the Stiefel manifold into densities over subsets of Euclidean space. We show how to deal with issues arising from the topology of the Stiefel manifold and how to inexpensively compute the change-of-measure terms. We introduce an auxiliary parameter approach that limits the impact of topological issues. We provide both analysis of our methods and numerical examples demonstrating the effectiveness of the approach. We also discuss how our Givens representation can be used to define general classes of distributions over the space of orthogonal matrices. We then give demonstrations on several examples showing how the Givens approach performs in practice in comparison with other methods.
引用
收藏
页码:639 / 666
页数:28
相关论文
共 50 条
  • [1] BAYESIAN NONPARAMETRIC INFERENCE ON THE STIEFEL MANIFOLD
    Lin, Lizhen
    Rao, Vinayak
    Dunson, David
    [J]. STATISTICA SINICA, 2017, 27 (02) : 535 - 553
  • [2] Bayesian filtering on the Stiefel manifold
    Tompkins, Frank
    Wolfe, Patrick J.
    [J]. 2007 2ND IEEE INTERNATIONAL WORKSHOP ON COMPUTATIONAL ADVANCES IN MULTI-SENSOR ADAPTIVE PROCESSING, 2007, : 161 - 164
  • [3] Variational Bayesian Orthogonal Nonnegative Matrix Factorization Over the Stiefel Manifold
    Rahiche, Abderrahmane
    Cheriet, Mohamed
    [J]. IEEE Transactions on Image Processing, 2022, 31 : 5543 - 5558
  • [4] Variational Bayesian Orthogonal Nonnegative Matrix Factorization Over the Stiefel Manifold
    Rahiche, Abderrahmane
    Cheriet, Mohamed
    [J]. IEEE TRANSACTIONS ON IMAGE PROCESSING, 2022, 31 : 5543 - 5558
  • [5] Interpretable domain adaptation via optimization over the Stiefel manifold
    Poelitz, Christian
    Duivesteijn, Wouter
    Morik, Katharina
    [J]. MACHINE LEARNING, 2016, 104 (2-3) : 315 - 336
  • [6] Interpretable domain adaptation via optimization over the Stiefel manifold
    Christian Pölitz
    Wouter Duivesteijn
    Katharina Morik
    [J]. Machine Learning, 2016, 104 : 315 - 336
  • [7] Quadratic programs over the Stiefel manifold
    Fan, JY
    Nie, PY
    [J]. OPERATIONS RESEARCH LETTERS, 2006, 34 (02) : 135 - 141
  • [8] Bayesian non-parametric inference for manifold based MoCap representation
    Natola, Fabrizio
    Ntouskos, Valsamis
    Sanzari, Marta
    Pirri, Fiora
    [J]. 2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, : 4606 - 4614
  • [9] Multivariate regression via Stiefel manifold constraints
    Bakir, GH
    Gretton, A
    Franz, M
    Schölkopf, B
    [J]. PATTERN RECOGNITION, 2004, 3175 : 262 - 269
  • [10] Fast MIMO Blind Detection via Modified MMA Approach Over the Stiefel Manifold
    Chen, Yantao
    Dong, Binhong
    Gao, Pengyu
    Xiong, Wenhui
    [J]. IEEE WIRELESS COMMUNICATIONS LETTERS, 2024, 13 (05) : 1483 - 1487