bibtype	C - Conference Paper (international conference)
ARLID	0360026
utime	20240111140756.3
mtime	20111108235959.9
title (primary) (eng)	Fast damped Gauss-Newton algorithm for sparse and nonnegative tensor factorization
specification	page_count 4 s.
serial	ARLID cav_un_epca*0363842 ISBN 978-1-4577-0539-7 title Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing 2011 page_num 1988-1991 publisher place Piscataway name IEEE year 2011
keyword	Multilinear models
keyword	canonical polyadic decomposition
keyword	nonegative tensor factorization
author (primary)	ARLID cav_un_auth*0274170 name1 Phan name2 A. H. country JP
author	ARLID cav_un_auth*0101212 name1 Tichavský name2 Petr full_dept (cz) Stochastická informatika full_dept Department of Stochastic Informatics department (cz) SI department SI institution UTIA-B full_dept Department of Stochastic Informatics fullinstit Ústav teorie informace a automatizace AV ČR, v. v. i.
author	ARLID cav_un_auth*0274171 name1 Cichocki name2 A. country JP
source	url http://library.utia.cas.cz/separaty/2011/SI/tichavsky-fast damped gauss-newton algorithm for nonnegative matrix factorization.pdf source_size 217 kB
cas_special	project project_id 1M0572 agency GA MŠk ARLID cav_un_auth*0001814 project project_id GA102/09/1278 agency GA ČR ARLID cav_un_auth*0253174 research CEZ:AV0Z10750506 abstract (eng) Alternating optimization algorithms for canonical polyadic decomposition (with/without nonnegative constraints) often accompany update rules with low computational cost, but could face problems of swamps, bottlenecks, and slow convergence. All-at-once algorithms can deal with such problems, but always demand significant temporary extra-storage, and high computational cost. In this paper, we propose an allat- once algorithmwith lowcomplexity for sparse and nonnegative tensor factorization based on the damped Gauss-Newton iteration. Especially, for low-rank approximations, the proposed algorithm avoids building up Hessians and gradients, reduces the computational cost dramatically. Moreover, we proposed selection strategies for regularization parameters. The proposed algorithm has been verified to overwhelmingly outperform “state-of-the-art” NTF algorithms for difficult benchmarks, and for real-world application such as clustering of the ORL face database. action ARLID cav_un_auth*0272319 name 2011 IEEE International Conference on Acoustics, Speech, and Signal Processing ICASSP 2011 place Praha dates 22.05.2011-27.05.2011 country CZ reportyear 2012 RIV BB num_of_auth 3 permalink http://hdl.handle.net/11104/0197677 arlyear 2011 mrcbU56 217 kB mrcbU63 cav_un_epca*0363842 Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing 2011 978-1-4577-0539-7 1988 1991 Piscataway IEEE 2011