Bayesian sparse partial least squares

Diego Vidaurre; Marcel A.J. Van Gerven; Concha Bielza; Pedro Larrañaga; Tom Heskes

doi:10.1162/NECO_a_00524

Bayesian sparse partial least squares

Diego Vidaurre^*, Marcel A.J. Van Gerven, Concha Bielza, Pedro Larrañaga, Tom Heskes

^*Corresponding author for this work

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Letter › peer-review

20 Citations (Scopus)

Abstract

Partial least squares (PLS) is a class of methods that makes use of a set of latent or unobserved variables to model the relation between (typically) two sets of input and output variables, respectively. Several flavors, depending on how the latent variables or components are computed, have been developed over the last years. In this letter, we propose a Bayesian formulation of PLS alongwith some extensions. In a nutshell, we provide sparsity at the input space level and an automatic estimation of the optimal number of latent components. We follow the variational approach to infer the parameter distributions.We have successfully tested the proposedmethods on a synthetic data benchmark and on electrocorticogram data associated with several motor outputs in monkeys.

Original language	English
Journal	Neural Computation
Volume	25
Issue	12
Pages (from-to)	3318-3339
Number of pages	22
ISSN	0899-7667
DOIs	https://doi.org/10.1162/NECO_a_00524
Publication status	Published - 19 Nov 2013
Externally published	Yes

Access to Document

10.1162/NECO_a_00524

Cite this

@article{aa9c188ae6c446dda5f3f699cc7b5cc6,

title = "Bayesian sparse partial least squares",

abstract = "Partial least squares (PLS) is a class of methods that makes use of a set of latent or unobserved variables to model the relation between (typically) two sets of input and output variables, respectively. Several flavors, depending on how the latent variables or components are computed, have been developed over the last years. In this letter, we propose a Bayesian formulation of PLS alongwith some extensions. In a nutshell, we provide sparsity at the input space level and an automatic estimation of the optimal number of latent components. We follow the variational approach to infer the parameter distributions.We have successfully tested the proposedmethods on a synthetic data benchmark and on electrocorticogram data associated with several motor outputs in monkeys.",

author = "Diego Vidaurre and {Van Gerven}, {Marcel A.J.} and Concha Bielza and Pedro Larra{\~n}aga and Tom Heskes",

year = "2013",

month = nov,

day = "19",

doi = "10.1162/NECO_a_00524",

language = "English",

volume = "25",

pages = "3318--3339",

journal = "Neural Computation",

issn = "0899-7667",

publisher = "MIT Press Journals",

number = "12",

}

TY - JOUR

T1 - Bayesian sparse partial least squares

AU - Vidaurre, Diego

AU - Van Gerven, Marcel A.J.

AU - Bielza, Concha

AU - Larrañaga, Pedro

AU - Heskes, Tom

PY - 2013/11/19

Y1 - 2013/11/19

N2 - Partial least squares (PLS) is a class of methods that makes use of a set of latent or unobserved variables to model the relation between (typically) two sets of input and output variables, respectively. Several flavors, depending on how the latent variables or components are computed, have been developed over the last years. In this letter, we propose a Bayesian formulation of PLS alongwith some extensions. In a nutshell, we provide sparsity at the input space level and an automatic estimation of the optimal number of latent components. We follow the variational approach to infer the parameter distributions.We have successfully tested the proposedmethods on a synthetic data benchmark and on electrocorticogram data associated with several motor outputs in monkeys.

AB - Partial least squares (PLS) is a class of methods that makes use of a set of latent or unobserved variables to model the relation between (typically) two sets of input and output variables, respectively. Several flavors, depending on how the latent variables or components are computed, have been developed over the last years. In this letter, we propose a Bayesian formulation of PLS alongwith some extensions. In a nutshell, we provide sparsity at the input space level and an automatic estimation of the optimal number of latent components. We follow the variational approach to infer the parameter distributions.We have successfully tested the proposedmethods on a synthetic data benchmark and on electrocorticogram data associated with several motor outputs in monkeys.

UR - http://www.scopus.com/inward/record.url?scp=84887564612&partnerID=8YFLogxK

U2 - 10.1162/NECO_a_00524

DO - 10.1162/NECO_a_00524

M3 - Letter

C2 - 24047319

AN - SCOPUS:84887564612

SN - 0899-7667

VL - 25

SP - 3318

EP - 3339

JO - Neural Computation

JF - Neural Computation

IS - 12

ER -

Bayesian sparse partial least squares

Abstract

Access to Document

Other files and links

Fingerprint

Cite this