Matches in SemOpenAlex for { <https://semopenalex.org/work/W3170845162> ?p ?o ?g. }
- W3170845162 abstract "Motivated by the recent work of Mangoubi and Vishnoi (STOC 2021), we propose a variant of the min-max optimization framework where the max-player is constrained to update the maximization variable in a greedy manner until it reaches a *first-order* stationary point. We present an algorithm that provably converges to an approximate local equilibrium for our framework from any initialization and for nonconvex-nonconcave loss functions. Compared to the second-order algorithm of Mangoubi and Vishnoi, whose iteration bound is polynomial in the dimension, our algorithm is first-order and its iteration bound is independent of dimension. We empirically evaluate our algorithm on challenging nonconvex-nonconcave test-functions and loss functions that arise in GAN training. Our algorithm converges on these test functions and, when used to train GANs on synthetic and real-world datasets, trains stably and avoids mode collapse." @default.
- W3170845162 created "2021-06-22" @default.
- W3170845162 creator A5006627464 @default.
- W3170845162 creator A5063089732 @default.
- W3170845162 creator A5067464936 @default.
- W3170845162 creator A5080439623 @default.
- W3170845162 date "2020-06-22" @default.
- W3170845162 modified "2023-10-01" @default.
- W3170845162 title "A Convergent and Dimension-Independent First-Order Algorithm for Min-Max Optimization" @default.
- W3170845162 cites W1516681311 @default.
- W3170845162 cites W1519983590 @default.
- W3170845162 cites W2009941369 @default.
- W3170845162 cites W2089559088 @default.
- W3170845162 cites W2099471712 @default.
- W3170845162 cites W2554314924 @default.
- W3170845162 cites W2604763608 @default.
- W3170845162 cites W2609037894 @default.
- W3170845162 cites W2619503996 @default.
- W3170845162 cites W2785464582 @default.
- W3170845162 cites W2894677249 @default.
- W3170845162 cites W2895628298 @default.
- W3170845162 cites W2907295878 @default.
- W3170845162 cites W2947761153 @default.
- W3170845162 cites W2948350736 @default.
- W3170845162 cites W2948517632 @default.
- W3170845162 cites W2962808998 @default.
- W3170845162 cites W2962837045 @default.
- W3170845162 cites W2962979365 @default.
- W3170845162 cites W2962981216 @default.
- W3170845162 cites W2963142510 @default.
- W3170845162 cites W2963147599 @default.
- W3170845162 cites W2963185411 @default.
- W3170845162 cites W2963326510 @default.
- W3170845162 cites W2963647223 @default.
- W3170845162 cites W2963865839 @default.
- W3170845162 cites W2963981733 @default.
- W3170845162 cites W2963989027 @default.
- W3170845162 cites W2964070557 @default.
- W3170845162 cites W2964106499 @default.
- W3170845162 cites W2964126461 @default.
- W3170845162 cites W2964201867 @default.
- W3170845162 cites W2964253222 @default.
- W3170845162 cites W2964334726 @default.
- W3170845162 cites W2970416783 @default.
- W3170845162 cites W2970552792 @default.
- W3170845162 cites W2979980060 @default.
- W3170845162 cites W2995291884 @default.
- W3170845162 cites W3016604587 @default.
- W3170845162 cites W3034862928 @default.
- W3170845162 cites W3035723062 @default.
- W3170845162 cites W3037593317 @default.
- W3170845162 cites W3105747145 @default.
- W3170845162 cites W3125405799 @default.
- W3170845162 cites W3131685225 @default.
- W3170845162 cites W3171002556 @default.
- W3170845162 hasPublicationYear "2020" @default.
- W3170845162 type Work @default.
- W3170845162 sameAs 3170845162 @default.
- W3170845162 citedByCount "0" @default.
- W3170845162 crossrefType "posted-content" @default.
- W3170845162 hasAuthorship W3170845162A5006627464 @default.
- W3170845162 hasAuthorship W3170845162A5063089732 @default.
- W3170845162 hasAuthorship W3170845162A5067464936 @default.
- W3170845162 hasAuthorship W3170845162A5080439623 @default.
- W3170845162 hasConcept C10138342 @default.
- W3170845162 hasConcept C11413529 @default.
- W3170845162 hasConcept C114466953 @default.
- W3170845162 hasConcept C114614502 @default.
- W3170845162 hasConcept C126255220 @default.
- W3170845162 hasConcept C134306372 @default.
- W3170845162 hasConcept C14036430 @default.
- W3170845162 hasConcept C162324750 @default.
- W3170845162 hasConcept C182306322 @default.
- W3170845162 hasConcept C182365436 @default.
- W3170845162 hasConcept C189237950 @default.
- W3170845162 hasConcept C199360897 @default.
- W3170845162 hasConcept C2776330181 @default.
- W3170845162 hasConcept C28826006 @default.
- W3170845162 hasConcept C33676613 @default.
- W3170845162 hasConcept C33923547 @default.
- W3170845162 hasConcept C41008148 @default.
- W3170845162 hasConcept C78458016 @default.
- W3170845162 hasConcept C86803240 @default.
- W3170845162 hasConceptScore W3170845162C10138342 @default.
- W3170845162 hasConceptScore W3170845162C11413529 @default.
- W3170845162 hasConceptScore W3170845162C114466953 @default.
- W3170845162 hasConceptScore W3170845162C114614502 @default.
- W3170845162 hasConceptScore W3170845162C126255220 @default.
- W3170845162 hasConceptScore W3170845162C134306372 @default.
- W3170845162 hasConceptScore W3170845162C14036430 @default.
- W3170845162 hasConceptScore W3170845162C162324750 @default.
- W3170845162 hasConceptScore W3170845162C182306322 @default.
- W3170845162 hasConceptScore W3170845162C182365436 @default.
- W3170845162 hasConceptScore W3170845162C189237950 @default.
- W3170845162 hasConceptScore W3170845162C199360897 @default.
- W3170845162 hasConceptScore W3170845162C2776330181 @default.
- W3170845162 hasConceptScore W3170845162C28826006 @default.
- W3170845162 hasConceptScore W3170845162C33676613 @default.
- W3170845162 hasConceptScore W3170845162C33923547 @default.
- W3170845162 hasConceptScore W3170845162C41008148 @default.