Diffusion LMS for multitask problems with overlapping hypothesis subspaces

Jie Chen; Cedric Richard; Alfred O. Hero; Ali H. Sayed

doi:10.1109/MLSP.2014.6958929

Diffusion LMS for multitask problems with overlapping hypothesis subspaces

Jie Chen
, Cedric Richard
, Alfred O. Hero
, Ali H. Sayed

University of Michigan, Ann Arbor
Université Côte d'Azur
University of California at Los Angeles

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

39 Scopus citations

Abstract

There are many important applications that are multitask-oriented in the sense that there are multiple optimum parameter vectors to be inferred simultaneously by networked agents. In this paper, we formulate an online multitask learning problem where node hypothesis spaces partly overlap. A cooperative algorithm based on diffusion adaptation is derived. Some results on its stability and convergence properties are also provided. Simulations are conducted to illustrate the theoretical results.

Original language	English
Title of host publication	2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014
Editors	Mamadou Mboup, Tulay Adali, Eric Moreau, Jan Larsen
Publisher	IEEE Computer Society
ISBN (Electronic)	9781479936946
DOIs	https://doi.org/10.1109/MLSP.2014.6958929
State	Published - 14 Nov 2014
Externally published	Yes
Event	24th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2014 - Reims, France Duration: 21 Sep 2014 → 24 Sep 2014

Publication series

Name	IEEE International Workshop on Machine Learning for Signal Processing, MLSP
ISSN (Print)	2161-0363
ISSN (Electronic)	2161-0371

Conference

Conference	24th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2014
Country/Territory	France
City	Reims
Period	21/09/14 → 24/09/14

Keywords

Multitask learning
collaborative processing
diffusion strategy
distributed optimization

Access to Document

10.1109/MLSP.2014.6958929

Cite this

Chen, J., Richard, C., Hero, A. O., & Sayed, A. H. (2014). Diffusion LMS for multitask problems with overlapping hypothesis subspaces. In M. Mboup, T. Adali, E. Moreau, & J. Larsen (Eds.), 2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014 Article 6958929 (IEEE International Workshop on Machine Learning for Signal Processing, MLSP). IEEE Computer Society. https://doi.org/10.1109/MLSP.2014.6958929

Chen, Jie ; Richard, Cedric ; Hero, Alfred O. et al. / Diffusion LMS for multitask problems with overlapping hypothesis subspaces. 2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014. editor / Mamadou Mboup ; Tulay Adali ; Eric Moreau ; Jan Larsen. IEEE Computer Society, 2014. (IEEE International Workshop on Machine Learning for Signal Processing, MLSP).

@inproceedings{1aa22781bda341829e6086cbc57d3969,

title = "Diffusion LMS for multitask problems with overlapping hypothesis subspaces",

abstract = "There are many important applications that are multitask-oriented in the sense that there are multiple optimum parameter vectors to be inferred simultaneously by networked agents. In this paper, we formulate an online multitask learning problem where node hypothesis spaces partly overlap. A cooperative algorithm based on diffusion adaptation is derived. Some results on its stability and convergence properties are also provided. Simulations are conducted to illustrate the theoretical results.",

keywords = "Multitask learning, collaborative processing, diffusion strategy, distributed optimization",

author = "Jie Chen and Cedric Richard and Hero, \{Alfred O.\} and Sayed, \{Ali H.\}",

note = "Publisher Copyright: {\textcopyright} 2014 IEEE.; 24th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2014 ; Conference date: 21-09-2014 Through 24-09-2014",

year = "2014",

month = nov,

day = "14",

doi = "10.1109/MLSP.2014.6958929",

language = "英语",

series = "IEEE International Workshop on Machine Learning for Signal Processing, MLSP",

publisher = "IEEE Computer Society",

editor = "Mamadou Mboup and Tulay Adali and Eric Moreau and Jan Larsen",

booktitle = "2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014",

}

Chen, J, Richard, C, Hero, AO & Sayed, AH 2014, Diffusion LMS for multitask problems with overlapping hypothesis subspaces. in M Mboup, T Adali, E Moreau & J Larsen (eds), 2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014., 6958929, IEEE International Workshop on Machine Learning for Signal Processing, MLSP, IEEE Computer Society, 24th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2014, Reims, France, 21/09/14. https://doi.org/10.1109/MLSP.2014.6958929

Diffusion LMS for multitask problems with overlapping hypothesis subspaces. / Chen, Jie; Richard, Cedric; Hero, Alfred O. et al.
2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014. ed. / Mamadou Mboup; Tulay Adali; Eric Moreau; Jan Larsen. IEEE Computer Society, 2014. 6958929 (IEEE International Workshop on Machine Learning for Signal Processing, MLSP).

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

TY - GEN

T1 - Diffusion LMS for multitask problems with overlapping hypothesis subspaces

AU - Chen, Jie

AU - Richard, Cedric

AU - Hero, Alfred O.

AU - Sayed, Ali H.

PY - 2014/11/14

Y1 - 2014/11/14

N2 - There are many important applications that are multitask-oriented in the sense that there are multiple optimum parameter vectors to be inferred simultaneously by networked agents. In this paper, we formulate an online multitask learning problem where node hypothesis spaces partly overlap. A cooperative algorithm based on diffusion adaptation is derived. Some results on its stability and convergence properties are also provided. Simulations are conducted to illustrate the theoretical results.

AB - There are many important applications that are multitask-oriented in the sense that there are multiple optimum parameter vectors to be inferred simultaneously by networked agents. In this paper, we formulate an online multitask learning problem where node hypothesis spaces partly overlap. A cooperative algorithm based on diffusion adaptation is derived. Some results on its stability and convergence properties are also provided. Simulations are conducted to illustrate the theoretical results.

KW - Multitask learning

KW - collaborative processing

KW - diffusion strategy

KW - distributed optimization

UR - https://www.scopus.com/pages/publications/84912569546

U2 - 10.1109/MLSP.2014.6958929

DO - 10.1109/MLSP.2014.6958929

M3 - 会议稿件

AN - SCOPUS:84912569546

T3 - IEEE International Workshop on Machine Learning for Signal Processing, MLSP

BT - 2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014

A2 - Mboup, Mamadou

A2 - Adali, Tulay

A2 - Moreau, Eric

A2 - Larsen, Jan

PB - IEEE Computer Society

T2 - 24th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2014

Y2 - 21 September 2014 through 24 September 2014

ER -

Chen J, Richard C, Hero AO, Sayed AH. Diffusion LMS for multitask problems with overlapping hypothesis subspaces. In Mboup M, Adali T, Moreau E, Larsen J, editors, 2014 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2014. IEEE Computer Society. 2014. 6958929. (IEEE International Workshop on Machine Learning for Signal Processing, MLSP). doi: 10.1109/MLSP.2014.6958929

Diffusion LMS for multitask problems with overlapping hypothesis subspaces

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this