• Corpus ID: 119148205

Refined Shapely-Folkman Lemma and Its Application in Duality Gap Estimation

@article{Bi2016RefinedSL,
  title={Refined Shapely-Folkman Lemma and Its Application in Duality Gap Estimation},
  author={Yingjie Bi and Ao Tang},
  journal={arXiv: Optimization and Control},
  year={2016}
}
  • Yingjie Bi, A. Tang
  • Published 18 October 2016
  • Computer Science
  • arXiv: Optimization and Control
Based on concepts like kth convex hull and finer characterization of nonconvexity of a function, we propose a refinement of the Shapley-Folkman lemma and derive a new estimate for the duality gap of nonconvex optimization problems with separable objective functions. We apply our result to a network flow problem and the dynamic spectrum management problem in communication systems as examples to demonstrate that the new bound can be qualitatively tighter than the existing ones. The idea is also… 

Figures from this paper

An Approximate Shapley-Folkman Theorem
The Shapley-Folkman theorem shows that Minkowski averages of uniformly bounded sets tend to be convex when the number of terms in the sum becomes much larger than the ambient dimension. In
The Nonconvex Geometry of Linear Inverse Problems
TLDR
The gaugep function is a simple generalization of the gauge function that can tightly control the sparsity of a statistical model within the learning alphabet and, perhaps surprisingly, draws further inspiration from the Burer-Monteiro factorization in computational mathematics.
Deep Neural Networks with Multi-Branch Architectures Are Intrinsically Less Non-Convex
TLDR
This work provides strong guarantees of this quantity for two classes of network architectures, for the neural networks with arbitrary activation functions, multi-branch architecture and a variant of hinge loss, and shows that the duality gap of both population and empirical risks shrinks to zero as the number of branches increases.
Deep Neural Networks with Multi-Branch Architectures Are Less Non-Convex
TLDR
It is shown that the duality gap of both population and empirical risks shrinks to zero as the number of branches increases, and sheds light on better understanding the power of over-parametrization where increasing the network width tends to make the loss surface less non-convex.

References

SHOWING 1-10 OF 10 REFERENCES
Estimates of the duality gap for large-scale separable nonconvex optimization problems
We derive some estimates of the duality gap for separable constrained optimization problems involving nonconvex, possibly discontinuous, objective functions, and nonconvex, possibly discrete,
Estimates of the Duality Gap in Nonconvex Optimization
TLDR
Approximating in this way problems where the criterion and constraints are given as integrals, it is shown that the duality gap vanishes.
Bounding duality gap for separable problems with linear constraints
TLDR
A randomized algorithm is proposed to solve the convexified problem of minimizing a sum of non-convex functions over a compact domain, subject to linear inequality and equality constraints, and gives a bound that is tight.
Duality Gap Estimation and Polynomial Time Approximation for Optimal Spectrum Management
TLDR
It is shown that the Lagrangian dual relaxation, combined with a linear programming scheme, can generate an epsiv-optimal solution for the continuous formulation of the spectrum management problem in polynomial time for any epsIV > 0.
Network utility maximization with path cardinality constraints
  • Yingjie Bi, C. Tan, A. Tang
  • Computer Science
    IEEE INFOCOM 2016 - The 35th Annual IEEE International Conference on Computer Communications
  • 2016
TLDR
A novel convex relaxation is provided, which leads to a randomized algorithm with performance guarantees, which enables distributed algorithm design and allows for performance estimation for nonconvex routing optimization problems that is significantly better than previous work based on the multipath routing relaxation.
Dual methods for nonconvex spectrum optimization of multicarrier systems
  • Wei Yu, R. Lui
  • Computer Science
    IEEE Transactions on Communications
  • 2006
TLDR
It is shown that under a certain condition called the time-sharing condition, the duality gap of the optimization problem is always zero, regardless of the convexity of the objective function, which leads to efficient numerical algorithms that solve the nonconvex problem in the dual domain.
Convex analysis and minimization algorithms
IX. Inner Construction of the Subdifferential.- X. Conjugacy in Convex Analysis.- XI. Approximate Subdifferentials of Convex Functions.- XII. Abstract Duality for Practitioners.- XIII. Methods of
Carathéodory-type results for the sums and unions of convex sets
Several Carathéodory-type results on extreme representations of sums and unions of finitely many closed convex sets or polyhedra in Rn in terms of their faces will be presented (based on a manuscript
Ak-extreme point is the limit ofk-exposed points
It is proved that the relative boundary of ak-dimensional intersection of a hyperplane and a compact convex body is contained in the closure of the union of all intersections of dimension lower thanp
Quasi-Equilibria in Markets with Non-Convex Preferences