TY - JOUR
T1 - Rate of convergence analysis of dual-based variables decomposition methods for strongly convex problems
AU - Beck, Amir
AU - Tetruashvili, Luba
AU - Vaisbourd, Yakov
AU - Shemtov, Ariel
N1 - Publisher Copyright:
© 2015 Elsevier B.V.
PY - 2016/1
Y1 - 2016/1
N2 - We consider the problem of minimizing the sum of a strongly convex function and a term comprising the sum of extended real-valued proper closed convex functions. We derive the primal representation of dual-based block descent methods and establish a relation between primal and dual rates of convergence, allowing to compute the efficiency estimates of different methods. We illustrate the effectiveness of the methods by numerical experiments on total variation-based denoising problems.
AB - We consider the problem of minimizing the sum of a strongly convex function and a term comprising the sum of extended real-valued proper closed convex functions. We derive the primal representation of dual-based block descent methods and establish a relation between primal and dual rates of convergence, allowing to compute the efficiency estimates of different methods. We illustrate the effectiveness of the methods by numerical experiments on total variation-based denoising problems.
KW - Block variables decomposition
KW - Dual based methods
KW - Total variation denoising
UR - http://www.scopus.com/inward/record.url?scp=84949058358&partnerID=8YFLogxK
U2 - 10.1016/j.orl.2015.11.007
DO - 10.1016/j.orl.2015.11.007
M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???
AN - SCOPUS:84949058358
SN - 0167-6377
VL - 44
SP - 61
EP - 66
JO - Operations Research Letters
JF - Operations Research Letters
IS - 1
ER -