Mathematics > Optimization and Control
[Submitted on 16 Nov 2012 (this version), latest version 11 Jun 2013 (v5)]
Title:Distance Majorization and Its Applications
View PDFAbstract:We study the problem of minimizing a convex function over an intersection of closed convex sets. We are particularly interested in cases where it is easy to compute the projection onto any given set, but nontrivial to project onto the intersection. Algorithms based on Newton's method such as the interior point method offer an alternative for small to medium-scale problems. However, modern applications in machine learning are posing problems with tens of thousands of parameters or more. We revisit this problem and propose an algorithm that scales well with dimensionality. Our proposal revolves around three ideas: the majorization-minimization (MM) principle, the classical penalty method for constrained optimization, and quasi-Newton acceleration of fixed-point algorithms. The performance of our distance majorization algorithms is demonstrated on some modern machine learning problems.
Submission history
From: Eric Chi [view email][v1] Fri, 16 Nov 2012 14:47:43 UTC (347 KB)
[v2] Wed, 9 Jan 2013 19:25:09 UTC (384 KB)
[v3] Mon, 11 Mar 2013 14:29:39 UTC (385 KB)
[v4] Thu, 23 May 2013 17:30:10 UTC (415 KB)
[v5] Tue, 11 Jun 2013 23:08:53 UTC (417 KB)
Current browse context:
math.OC
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.