Stochastic Projective Splitting: Solving Saddle-Point Problems with Multiple Regularizers
Not on the map. Why?
Abstract (arXiv)
We present a new, stochastic variant of the projective splitting (PS) family of algorithms for monotone inclusion problems. It can solve min-max and noncooperative game formulations arising in applications such as robust ML without the convergence issues associated with gradient descent-ascent, the current de facto standard approach in such situations. Our proposal is the first version of PS able to use stochastic (as opposed to deterministic) gradient oracles. It is also the first stochastic method that can solve min-max games while easily handling multiple constraints and nonsmooth regularizers via projection and proximal operators. We close with numerical experiments on a distributionally robust sparse logistic regression problem.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 36 of its 70 references that are in the map. First posted on arXiv 24 Jun 2021.
Builds on
36 of 70 references.
References · 36
- 2018Which Training Methods for GANs do actually Converge?cited by 672
- 2018Mitigating Unwanted Biases with Adversarial Learningcited by 663
- 2017Certifying Some Distributional Robustness with Principled Adversarial Trainingcited by 333
- 2019On Gradient Descent Ascent for Nonconvex-Concave Minimax Problemscited by 266
- 2017Training GANs with Optimismcited by 217
- 2015Censoring Representations with an Adversarycited by 191
- 2019Wasserstein Distributionally Robust Optimization: Theory and Applications in Machine Learningcited by 169
- 2018A Variational Inequality Perspective on Generative Adversarial Networkscited by 144
- 2018Optimistic mirror descent in saddle-point problems: Going the extra (gradient) milecited by 132
- 2015Distributionally Robust Logistic Regressioncited by 124
- 2018The Mechanics of n-Player Differentiable Gamescited by 118
- 2017The Numerics of GANscited by 104
24 more
- 2019On the convergence of single-call stochastic extra-gradient methodscited by 88
- 2018Achieving Fairness through Adversarial Learning: an Application to Recidivism Predictioncited by 77
- 2017Gradient descent GAN optimization is locally stablecited by 65
- 2020Explore Aggressively, Update Conservatively: Stochastic Extragradient Methods with Variable Stepsize Scalingcited by 49
- 2018A Forward-Backward Splitting Method for Monotone Inclusions Without Cocoercivitycited by 35
- 2015A Three-Operator Splitting Scheme and its Optimization Applicationscited by 34
- 2017Context-Aware Generative Adversarial Privacycited by 27
- 2019Improved Adversarial Learning for Fair Classificationcited by 25
- 2014Optimal stochastic extragradient schemes for pseudomonotone stochastic variational inequality problems and their variantscited by 19
- 2021Fast Distributionally Robust Learning with Variance Reduced Min-Max Optimizationcited by 19
- 2020Taming GANs with Lookahead-Minmaxcited by 16
- 2020Two steps at a time -- taking GAN training in stride with Tseng's methodcited by 8
- 2018Adaptive Three Operator Splittingcited by 7
- 2018Rare Feature Selection in High Dimensionscited by 5
- 2019Forward-backward-forward methods with variance reduction for stochastic variational inequalitiescited by 5
- 2015Asynchronous Block-Iterative Primal-Dual Decomposition Methods for Monotone Inclusionscited by 4
- 2017Stochastic Three-Composite Convex Minimizationcited by 3
- 2021Generative Minimization Networks: Training GANs Without Competitioncited by 3
- 2013Solving Coupled Composite Monotone Inclusions by Successive Fej'er Approximations of Their Kuhn-Tucker Setcited by 1
- 2018Projective Splitting with Forward Steps: Asynchronous and Block-Iterative Operator Splittingcited by 1
- 2018Convergence Rates for Projective Splittingcited by 1
- 2018Projective Splitting with Forward Steps only Requires Continuitycited by 1
- 2019Single-Forward-Step Projective Splitting: Exploiting Cocoercivitycited by 1
- 2021Convergence analysis of the stochastic reflected forward-backward splitting algorithmcited by 1
Built on by
Citing papers per year: 2023: 1.
1 paper.