• Corpus ID: 219708662

The Landscape of Nonconvex-Nonconcave Minimax Optimization

  title={The Landscape of Nonconvex-Nonconcave Minimax Optimization},
  author={Benjamin Grimmer and Haihao Lu and Pratik Worah and Vahab S. Mirrokni},
Minimax optimization has become a central tool for modern machine learning with applications in robust optimization, game theory and training GANs. These applications are often nonconvex-nonconcave, but the existing theory is unable to identify and deal with the fundamental difficulties posed by nonconvex-nonconcave structures. We break this historical barrier by identifying three regions of nonconvex-nonconcave bilinear minimax problems and characterizing their different solution paths. For… 

Figures from this paper

A Unified Single-loop Alternating Gradient Projection Algorithm for Nonconvex-Concave and Convex-Nonconcave Minimax Problems
To the best of the authors' knowledge, this is the first time that a simple and unified single-loop algorithm is developed for solving both nonconvex-( strongly) concave and (strongly) convex-nonconcave minimax problems.
Finding First-Order Nash Equilibria of Zero-Sum Games with the Regularized Nikaido-Isoda Function
This work proposes an algorithm for finding the FNEs of a two-player zero-sum game, in which the local cost functions can be non-convex, and the players only have access to local stochastic gradients.
An $$O(s^r)$$-resolution ODE framework for understanding discrete-time algorithms and applications to the linear convergence of minimax problems
The proposed O(s)-linear-convergence conditions not only unify the known scenarios when PPM and EGM have linear convergence, but also showcase that these two algorithms exhibit linear convergence in much broader contexts, including when solving a class of nonconvex-nonconcave minimax problems.
The limits of min-max optimization algorithms: convergence to spurious non-critical sets
This paper characterize the convergence properties of a wide class of zeroth-, first-, and (scalable) second-order methods in non-convex/non-concave problems and shows that these state-of-the-art min-max optimization algorithms may converge with arbitrarily high probability to attractors that are in no way min- max optimal or even stationary.


Global Convergence and Variance-Reduced Optimization for a Class of Nonconvex-Nonconcave Minimax Problems
This work shows that for a subclass of nonconvex-nonconcave objectives satisfying a so-called two-sided Polyak-Łojasiewicz inequality, the alternating gradient descent ascent (AGDA) algorithm converges globally at a linear rate and the stochastic AGDA achieves a sublinear rate.
Proximally Guided Stochastic Subgradient Method for Nonsmooth, Nonconvex Problems
A simple proof that the proposed algorithm converges at the same rate as the stochastic gradient method for smooth nonconvex problems is presented, which appears to be the first convergence rate analysis of a Stochastic subgradient method for the class of weakly convex functions.
Minmax Optimization: Stable Limit Points of Gradient Descent Ascent are Locally Optimal
It is shown that as the ratio of the ascent step size to the descent step size goes to infinity, stable limit points of GDA are exactly local minmax points up to degenerate points, demonstrating that all stable limitpoints of G DA have a game-theoretic meaning for minmax problems.
Solving a Class of Non-Convex Min-Max Games Using Iterative First Order Methods
This paper proposes a multi-step gradient descent-ascent algorithm that finds an \varepsilon--first order stationary point of the game in \widetilde O(\varpsilon^{-3.5}) iterations, which is the best known rate in the literature.
Non-Convex Min-Max Optimization: Provable Algorithms and Applications in Machine Learning
This paper proposes a proximally guided stochastic subgradient method and a proxIMally guided Stochastic variance-reduced method for expected and finite-sum saddle-point problems, respectively and establishes the computation complexities of both methods for finding a nearly stationary point of the corresponding minimization problem.
Solving Weakly-Convex-Weakly-Concave Saddle-Point Problems as Successive Strongly Monotone Variational Inequalities
This work proposes an algorithmic framework motivated by the inexact proximal point method, which solves the weakly monotone variational inequality corresponding to the original min-max problem by approximately solving a sequence of strongly monot one variational inequalities constructed by adding a strongly monOTone mapping to theOriginal gradient mapping.
On minimax robustness: A general approach and applications
It is shown that if the performance functional and the uncertainty set are convex then a certain type of regularity condition on the functional is sufficient to ensure that the optimal strategy for a least favorable element of the uncertaintySet is minimax robust.
Stochastic model-based minimization of weakly convex functions
This work shows that under weak-convexity and Lipschitz conditions, the algorithm drives the expected norm of the gradient of the Moreau envelope to zero at the rate of $O(k^{-1/4})$.
The Limit Points of (Optimistic) Gradient Descent in Min-Max Optimization
This work characterize the limit points of two basic first order methods, namely Gradient Descent/Ascent (GDA) and Optimistic Gradients Descent Ascent (OGDA), and shows that both dynamics avoid unstable critical points for almost all initializations.
On Gradient Descent Ascent for Nonconvex-Concave Minimax Problems
This is the first nonasymptotic analysis for two-time-scale GDA in this setting, shedding light on its superior practical performance in training generative adversarial networks (GANs) and other real applications.