Matches in SemOpenAlex for { <https://semopenalex.org/work/W836608889> ?p ?o ?g. }
- W836608889 endingPage "1602" @default.
- W836608889 startingPage "1594" @default.
- W836608889 abstract "Stochastic convex optimization is a basic and well studied primitive in machine learning. It is well known that convex and Lipschitz functions can be minimized efficiently using Stochastic Gradient Descent (SGD).The Normalized Gradient Descent (NGD) algorithm, is an adaptation of Gradient Descent, which updates according to the direction of the gradients, rather than the gradients themselves. In this paper we analyze a stochastic version of NGD and prove its convergence to a global minimum for a wider class of functions: we require the functions to be quasi-convex and locally-Lipschitz. Quasi-convexity broadens the concept of unimodality to multidimensions and allows for certain types of saddle points, which are a known hurdle for first-order optimization methods such as gradient descent. Locally-Lipschitz functions are only required to be Lipschitz in a small region around the optimum. This assumption circumvents gradient explosion, which is another known hurdle for gradient descent variants. Interestingly, unlike the vanilla SGD algorithm, the stochastic normalized gradient descent algorithm provably requires a minimal minibatch size." @default.
- W836608889 created "2016-06-24" @default.
- W836608889 creator A5022424856 @default.
- W836608889 creator A5024431603 @default.
- W836608889 creator A5078263052 @default.
- W836608889 date "2015-12-07" @default.
- W836608889 modified "2023-09-23" @default.
- W836608889 title "Beyond convexity: stochastic quasi-convex optimization" @default.
- W836608889 cites W104184427 @default.
- W836608889 cites W1408639475 @default.
- W836608889 cites W1560025682 @default.
- W836608889 cites W1582743214 @default.
- W836608889 cites W1589928278 @default.
- W836608889 cites W173421169 @default.
- W836608889 cites W178986290 @default.
- W836608889 cites W1815076433 @default.
- W836608889 cites W1966986674 @default.
- W836608889 cites W2015960758 @default.
- W836608889 cites W2042799324 @default.
- W836608889 cites W2048395921 @default.
- W836608889 cites W2072128103 @default.
- W836608889 cites W2087228593 @default.
- W836608889 cites W2101762657 @default.
- W836608889 cites W2107878631 @default.
- W836608889 cites W2296319761 @default.
- W836608889 cites W2754843966 @default.
- W836608889 hasPublicationYear "2015" @default.
- W836608889 type Work @default.
- W836608889 sameAs 836608889 @default.
- W836608889 citedByCount "36" @default.
- W836608889 countsByYear W8366088892015 @default.
- W836608889 countsByYear W8366088892016 @default.
- W836608889 countsByYear W8366088892018 @default.
- W836608889 countsByYear W8366088892019 @default.
- W836608889 countsByYear W8366088892020 @default.
- W836608889 countsByYear W8366088892021 @default.
- W836608889 countsByYear W8366088892023 @default.
- W836608889 crossrefType "proceedings-article" @default.
- W836608889 hasAuthorship W836608889A5022424856 @default.
- W836608889 hasAuthorship W836608889A5024431603 @default.
- W836608889 hasAuthorship W836608889A5078263052 @default.
- W836608889 hasConcept C106159729 @default.
- W836608889 hasConcept C112680207 @default.
- W836608889 hasConcept C115680565 @default.
- W836608889 hasConcept C126255220 @default.
- W836608889 hasConcept C134306372 @default.
- W836608889 hasConcept C145446738 @default.
- W836608889 hasConcept C153258448 @default.
- W836608889 hasConcept C154945302 @default.
- W836608889 hasConcept C157972887 @default.
- W836608889 hasConcept C162324750 @default.
- W836608889 hasConcept C194387892 @default.
- W836608889 hasConcept C206688291 @default.
- W836608889 hasConcept C22324862 @default.
- W836608889 hasConcept C2524010 @default.
- W836608889 hasConcept C2681867 @default.
- W836608889 hasConcept C2777303404 @default.
- W836608889 hasConcept C28826006 @default.
- W836608889 hasConcept C33923547 @default.
- W836608889 hasConcept C41008148 @default.
- W836608889 hasConcept C50522688 @default.
- W836608889 hasConcept C50644808 @default.
- W836608889 hasConcept C72134830 @default.
- W836608889 hasConceptScore W836608889C106159729 @default.
- W836608889 hasConceptScore W836608889C112680207 @default.
- W836608889 hasConceptScore W836608889C115680565 @default.
- W836608889 hasConceptScore W836608889C126255220 @default.
- W836608889 hasConceptScore W836608889C134306372 @default.
- W836608889 hasConceptScore W836608889C145446738 @default.
- W836608889 hasConceptScore W836608889C153258448 @default.
- W836608889 hasConceptScore W836608889C154945302 @default.
- W836608889 hasConceptScore W836608889C157972887 @default.
- W836608889 hasConceptScore W836608889C162324750 @default.
- W836608889 hasConceptScore W836608889C194387892 @default.
- W836608889 hasConceptScore W836608889C206688291 @default.
- W836608889 hasConceptScore W836608889C22324862 @default.
- W836608889 hasConceptScore W836608889C2524010 @default.
- W836608889 hasConceptScore W836608889C2681867 @default.
- W836608889 hasConceptScore W836608889C2777303404 @default.
- W836608889 hasConceptScore W836608889C28826006 @default.
- W836608889 hasConceptScore W836608889C33923547 @default.
- W836608889 hasConceptScore W836608889C41008148 @default.
- W836608889 hasConceptScore W836608889C50522688 @default.
- W836608889 hasConceptScore W836608889C50644808 @default.
- W836608889 hasConceptScore W836608889C72134830 @default.
- W836608889 hasOpenAccess W836608889 @default.
- W836608889 hasRelatedWork W1505731132 @default.
- W836608889 hasRelatedWork W1815076433 @default.
- W836608889 hasRelatedWork W1988720110 @default.
- W836608889 hasRelatedWork W1992208280 @default.
- W836608889 hasRelatedWork W1994616650 @default.
- W836608889 hasRelatedWork W2009941369 @default.
- W836608889 hasRelatedWork W2107438106 @default.
- W836608889 hasRelatedWork W2146502635 @default.
- W836608889 hasRelatedWork W2194775991 @default.
- W836608889 hasRelatedWork W2296319761 @default.
- W836608889 hasRelatedWork W2561193401 @default.
- W836608889 hasRelatedWork W2622263826 @default.