How effective is the nuclear norm heuristic in solving data approximation problems?
How effective is the nuclear norm heuristic in solving data approximation problems?
The question in the title is answered empirically by solving instances of three classical problems: fitting a straight line to data, fitting a real exponent to data, and system identification in the errors-in-variables setting. The results show that the nuclear norm heuristic performs worse than alternative problem dependant methods---ordinary and total least squares, Kung's method, and subspace identification. In the line fitting and exponential fitting problems, the globally optimal solution is known analytically, so that the suboptimality of the heuristic methods is quantified.
Markovsky, Ivan
7d632d37-2100-41be-a4ff-90b92752212c
July 2012
Markovsky, Ivan
7d632d37-2100-41be-a4ff-90b92752212c
Markovsky, Ivan
(2012)
How effective is the nuclear norm heuristic in solving data approximation problems?
16th IFAC Symposium on System Identification (Sysid 2012), Brussels, Belgium.
11 - 13 Jul 2012.
6 pp
.
Record type:
Conference or Workshop Item
(Other)
Abstract
The question in the title is answered empirically by solving instances of three classical problems: fitting a straight line to data, fitting a real exponent to data, and system identification in the errors-in-variables setting. The results show that the nuclear norm heuristic performs worse than alternative problem dependant methods---ordinary and total least squares, Kung's method, and subspace identification. In the line fitting and exponential fitting problems, the globally optimal solution is known analytically, so that the suboptimality of the heuristic methods is quantified.
Archive
nucnrm-sysid-code.tar
- Other
Text
nucnrm-sysid.pdf
- Other
More information
Submitted date: March 2012
Published date: July 2012
Venue - Dates:
16th IFAC Symposium on System Identification (Sysid 2012), Brussels, Belgium, 2012-07-11 - 2012-07-13
Organisations:
Southampton Wireless Group
Identifiers
Local EPrints ID: 336088
URI: http://eprints.soton.ac.uk/id/eprint/336088
PURE UUID: a63e0acd-fdce-407e-83fb-eb5083567b4f
Catalogue record
Date deposited: 14 Mar 2012 17:16
Last modified: 14 Mar 2024 10:39
Export record
Contributors
Author:
Ivan Markovsky
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics