ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 1708.02105
114
51
v1v2v3 (latest)

Linear Convergence of a Frank-Wolfe Type Algorithm over Trace-Norm Balls

7 August 2017
Zeyuan Allen-Zhu
Elad Hazan
Wei Hu
Yuanzhi Li
ArXiv (abs)PDFHTML
Abstract

We propose a rank-kkk variant of the classical Frank-Wolfe algorithm to solve convex optimization over a trace-norm ball. Our algorithm replaces the top singular-vector computation (111-SVD) in Frank-Wolfe with a top-kkk singular-vector computation (kkk-SVD), which can be done by repeatedly applying 111-SVD kkk times. Alternatively, our algorithm can be viewed as a rank-kkk restricted version of projected gradient descent. We show that our algorithm has a linear convergence rate when the objective function is smooth and strongly convex, and the optimal solution has rank at most kkk. This improves the convergence rate and the total time complexity of the Frank-Wolfe method and its variants.

View on arXiv
Comments on this paper