Running MAP inference on million node graphical models: A high performance computing perspective

Chen Jin; Qiang Fu; Huahua Wang; William Hendrix; Zhengzhang Chen; Ankit Agrawal; Arindam Banerjee; Alok Choudhary

doi:10.1109/CCGrid.2015.35

Running MAP inference on million node graphical models: A high performance computing perspective

Chen Jin, Qiang Fu, Huahua Wang, William Hendrix, Zhengzhang Chen, Ankit Agrawal, Arindam Banerjee, Alok Choudhary

Computer Science and Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1 Scopus citations

Abstract

An important problem in discrete graphical models is the maximum a posterior (MAP) inference problem. Recent research has been focusing on the development of parallel MAP inference algorithm, which scales to graphical models of millions of nodes. In this paper, we introduce a parallel implementation of the recently proposed Bethe-ADMM algorithm using Message Passing Interface (MPI), which allows us to fully utilize the computing power provided by the modern supercomputers with thousands of cores. Experimental results demonstrate that for a broad class of problems, our parallel implementation of Bethe-ADMM scales almost linearly even with thousands of cores.

Original language	English (US)
Title of host publication	Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	565-575
Number of pages	11
ISBN (Electronic)	9781479980062
DOIs	https://doi.org/10.1109/CCGrid.2015.35
State	Published - Jul 7 2015
Event	15th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015 - Shenzhen, China Duration: May 4 2015 → May 7 2015

Publication series

Name	Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015

Other

Other	15th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015
Country/Territory	China
City	Shenzhen
Period	5/4/15 → 5/7/15

Bibliographical note

Publisher Copyright:
© 2015 IEEE.

Keywords

Alternating direction method of multipliers
Markov random field
Maximum a posteriori inference
Message passing interface

Access

10.1109/CCGrid.2015.35

OpenUrl availability

Full text

Cite this

Jin, C., Fu, Q., Wang, H., Hendrix, W., Chen, Z., Agrawal, A., Banerjee, A., & Choudhary, A. (2015). Running MAP inference on million node graphical models: A high performance computing perspective. In Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015 (pp. 565-575). Article 7152522 (Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CCGrid.2015.35

Running MAP inference on million node graphical models: A high performance computing perspective. / Jin, Chen; Fu, Qiang; Wang, Huahua et al.
Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015. Institute of Electrical and Electronics Engineers Inc., 2015. p. 565-575 7152522 (Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Jin, C, Fu, Q, Wang, H, Hendrix, W, Chen, Z, Agrawal, A, Banerjee, A & Choudhary, A 2015, Running MAP inference on million node graphical models: A high performance computing perspective. in Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015., 7152522, Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015, Institute of Electrical and Electronics Engineers Inc., pp. 565-575, 15th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015, Shenzhen, China, 5/4/15. https://doi.org/10.1109/CCGrid.2015.35

Jin C, Fu Q, Wang H, Hendrix W, Chen Z, Agrawal A et al. Running MAP inference on million node graphical models: A high performance computing perspective. In Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015. Institute of Electrical and Electronics Engineers Inc. 2015. p. 565-575. 7152522. (Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015). doi: 10.1109/CCGrid.2015.35

Jin, Chen ; Fu, Qiang ; Wang, Huahua et al. / Running MAP inference on million node graphical models : A high performance computing perspective. Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015. Institute of Electrical and Electronics Engineers Inc., 2015. pp. 565-575 (Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015).

@inproceedings{aa16d1cd07ff4ad2ac7576ac19cd862b,

title = "Running MAP inference on million node graphical models: A high performance computing perspective",

abstract = "An important problem in discrete graphical models is the maximum a posterior (MAP) inference problem. Recent research has been focusing on the development of parallel MAP inference algorithm, which scales to graphical models of millions of nodes. In this paper, we introduce a parallel implementation of the recently proposed Bethe-ADMM algorithm using Message Passing Interface (MPI), which allows us to fully utilize the computing power provided by the modern supercomputers with thousands of cores. Experimental results demonstrate that for a broad class of problems, our parallel implementation of Bethe-ADMM scales almost linearly even with thousands of cores.",

keywords = "Alternating direction method of multipliers, Markov random field, Maximum a posteriori inference, Message passing interface",

author = "Chen Jin and Qiang Fu and Huahua Wang and William Hendrix and Zhengzhang Chen and Ankit Agrawal and Arindam Banerjee and Alok Choudhary",

note = "Publisher Copyright: {\textcopyright} 2015 IEEE.; 15th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015 ; Conference date: 04-05-2015 Through 07-05-2015",

year = "2015",

month = jul,

day = "7",

doi = "10.1109/CCGrid.2015.35",

language = "English (US)",

series = "Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "565--575",

booktitle = "Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015",

}

TY - GEN

T1 - Running MAP inference on million node graphical models

T2 - 15th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015

AU - Jin, Chen

AU - Fu, Qiang

AU - Wang, Huahua

AU - Hendrix, William

AU - Chen, Zhengzhang

AU - Agrawal, Ankit

AU - Banerjee, Arindam

AU - Choudhary, Alok

PY - 2015/7/7

Y1 - 2015/7/7

N2 - An important problem in discrete graphical models is the maximum a posterior (MAP) inference problem. Recent research has been focusing on the development of parallel MAP inference algorithm, which scales to graphical models of millions of nodes. In this paper, we introduce a parallel implementation of the recently proposed Bethe-ADMM algorithm using Message Passing Interface (MPI), which allows us to fully utilize the computing power provided by the modern supercomputers with thousands of cores. Experimental results demonstrate that for a broad class of problems, our parallel implementation of Bethe-ADMM scales almost linearly even with thousands of cores.

AB - An important problem in discrete graphical models is the maximum a posterior (MAP) inference problem. Recent research has been focusing on the development of parallel MAP inference algorithm, which scales to graphical models of millions of nodes. In this paper, we introduce a parallel implementation of the recently proposed Bethe-ADMM algorithm using Message Passing Interface (MPI), which allows us to fully utilize the computing power provided by the modern supercomputers with thousands of cores. Experimental results demonstrate that for a broad class of problems, our parallel implementation of Bethe-ADMM scales almost linearly even with thousands of cores.

KW - Alternating direction method of multipliers

KW - Markov random field

KW - Maximum a posteriori inference

KW - Message passing interface

UR - http://www.scopus.com/inward/record.url?scp=84941248016&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84941248016&partnerID=8YFLogxK

U2 - 10.1109/CCGrid.2015.35

DO - 10.1109/CCGrid.2015.35

M3 - Conference contribution

AN - SCOPUS:84941248016

T3 - Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015

SP - 565

EP - 575

BT - Proceedings - 2015 IEEE/ACM 15th International Symposium on Cluster, Cloud, and Grid Computing, CCGrid 2015

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 4 May 2015 through 7 May 2015

ER -

Running MAP inference on million node graphical models: A high performance computing perspective

Abstract

Publication series

Other

Bibliographical note

Keywords

Access

OpenUrl availability

Other files and links

Fingerprint

Cite this