On approximations of the beta process in latent feature models: Point processes approach

被引：6

作者：

Al Labadi L. ^{[1
]}

Zarepour M. ^{[2
]}

机构：

[1] Department of Statistical Sciences, University of Toronto, Toronto, M5S 3G3, ON

[2] Department of Mathematics and Statistics, University of Ottawa, Ottawa, K1N 6N5, ON

来源：

Sankhya A | 2018年 / 80卷 / 1期

关键词：

Beta process; Ferguson and Klass representation; Finite dimensional approximation; Latent feature models; Simulation;

D O I：

10.1007/s13171-017-0103-9

中图分类号：

学科分类号：

摘要：

In recent times, the beta process has been widely used as a nonparametric prior for different models in machine learning, including latent feature models. In this paper, we prove the asymptotic consistency of the finite dimensional approximation of the beta process due to Paisley and Carin (2009). In particular, we show that this finite approximation converges in distribution to the Ferguson and Klass representation of the beta process. We implement this approximation to derive asymptotic properties of functionals of the finite dimensional beta process. In addition, we derive an almost sure approximation of the beta process. This new approximation provides a direct method to efficiently simulate the beta process. A simulated example, illustrating the work of the method and comparing its performance to several existing algorithms, is also included. © 2017, Indian Statistical Institute.

引用

页码：59 / 79

页数：20

共 38 条

[1]

Abramowitz M., Stegun I.A., Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, 10Th, (1972)

[2]

Al Labadi L., Zarepour M., On asymptotic properties and almost sure approximation of the norMalized inverse-Gaussian process, Bayesian Anal, 8, 3, pp. 553-568, (2013)

[3]

Al Labadi L., Zarepour M., A Bayesian nonparametric goodness of fit test for right censored data based on approximate samples from the beta-Stacy process, Canad. J. Statist, 41, 3, pp. 466-487, (2013)

[4]

Al Labadi L., Zarepour M., On Simulations from the Two-Parameter Poisson-Dirichlet Process and the NorMalized Inverse-Gaussian Process, Sankhya A, 76, pp. 158-176, (2014)

[5]

Al Labadi L., Zarepour M., Goodness of fit tests based on the distance between the Dirichlet process and its base measure, J. Nonparametr. Stat, 26, pp. 341-357, (2014)

[6]

Billingsley P., Convergence of Probability Measures, (1999)

[7]

Broderick T., Jordan M.I., Pitman J., Beta processes, stick-breaking, and power laws, Bayesian Anal, 7, pp. 439-476, (2012)

[8]

Chen B., Paisley J., Carin L., Sparse linear regression with beta process priors, International Conference on Acoustics, Speech and Signal Processing, (2010)

[9]

Chen B., Chen M., Paisley J., Zaas A., Woods C., Ginsburg G.S., Hero A., Lucas J., Dunson D., Carin L., Bayesian inference of the number of factors in gene-expression analysis: Application to human virus challenge studies, BMC Bioinformatics, 11, (2010)

[10]

Damien P., Laud P., Smith A.F.M., Approximate random variate generation from infinitely divisible distributions with applications to Bayesian inference, J. R. Stat. Soc. Ser. B Stat. Methodol, 57, pp. 547-563, (1995)

← 1 2 3 4 →