Matches in SemOpenAlex for { <https://semopenalex.org/work/W1788809966> ?p ?o ?g. }
Showing items 1 to 93 of
93
with 100 items per page.
- W1788809966 endingPage "533" @default.
- W1788809966 startingPage "525" @default.
- W1788809966 abstract "We prove that many mirror descent algorithms for online convex optimization (such as online gradient descent) have an equivalent interpretation as follow-the-regularizedleader (FTRL) algorithms. This observation makes the relationships between many commonly used algorithms explicit, and provides theoretical insight on previous experimental observations. In particular, even though the FOBOS composite mirror descent algorithm handles L1 regularization explicitly, it has been observed that the FTRL-style Regularized Dual Averaging (RDA) algorithm is even more effective at producing sparsity. Our results demonstrate that the key difference between these algorithms is how they handle the cumulative L1 penalty. While FOBOS handles the L1 term exactly on any given update, we show that it is effectively using subgradient approximations to the L1 penalty from previous rounds, leading to less sparsity than RDA, which handles the cumulative penalty in closed form. The FTRL-Proximal algorithm, which we introduce, can be seen as a hybrid of these two algorithms, and significantly outperforms both on a large, realworld dataset." @default.
- W1788809966 created "2016-06-24" @default.
- W1788809966 creator A5040311844 @default.
- W1788809966 date "2011-06-14" @default.
- W1788809966 modified "2023-09-26" @default.
- W1788809966 title "Follow-the-Regularized-Leader and Mirror Descent: Equivalence Theorems and L1 Regularization" @default.
- W1788809966 cites W131378802 @default.
- W1788809966 cites W1518461880 @default.
- W1788809966 cites W2096199223 @default.
- W1788809966 cites W2106129230 @default.
- W1788809966 cites W2139759436 @default.
- W1788809966 cites W2146502635 @default.
- W1788809966 cites W2148825261 @default.
- W1788809966 cites W2296319761 @default.
- W1788809966 cites W60189517 @default.
- W1788809966 cites W86575217 @default.
- W1788809966 hasPublicationYear "2011" @default.
- W1788809966 type Work @default.
- W1788809966 sameAs 1788809966 @default.
- W1788809966 citedByCount "75" @default.
- W1788809966 countsByYear W17888099662012 @default.
- W1788809966 countsByYear W17888099662013 @default.
- W1788809966 countsByYear W17888099662014 @default.
- W1788809966 countsByYear W17888099662015 @default.
- W1788809966 countsByYear W17888099662016 @default.
- W1788809966 countsByYear W17888099662017 @default.
- W1788809966 countsByYear W17888099662018 @default.
- W1788809966 countsByYear W17888099662019 @default.
- W1788809966 countsByYear W17888099662020 @default.
- W1788809966 countsByYear W17888099662021 @default.
- W1788809966 crossrefType "proceedings-article" @default.
- W1788809966 hasAuthorship W1788809966A5040311844 @default.
- W1788809966 hasConcept C11413529 @default.
- W1788809966 hasConcept C118615104 @default.
- W1788809966 hasConcept C126255220 @default.
- W1788809966 hasConcept C127413603 @default.
- W1788809966 hasConcept C146978453 @default.
- W1788809966 hasConcept C153258448 @default.
- W1788809966 hasConcept C154945302 @default.
- W1788809966 hasConcept C158968445 @default.
- W1788809966 hasConcept C196921405 @default.
- W1788809966 hasConcept C206688291 @default.
- W1788809966 hasConcept C2776135515 @default.
- W1788809966 hasConcept C2776637919 @default.
- W1788809966 hasConcept C2780069185 @default.
- W1788809966 hasConcept C33923547 @default.
- W1788809966 hasConcept C41008148 @default.
- W1788809966 hasConcept C50644808 @default.
- W1788809966 hasConceptScore W1788809966C11413529 @default.
- W1788809966 hasConceptScore W1788809966C118615104 @default.
- W1788809966 hasConceptScore W1788809966C126255220 @default.
- W1788809966 hasConceptScore W1788809966C127413603 @default.
- W1788809966 hasConceptScore W1788809966C146978453 @default.
- W1788809966 hasConceptScore W1788809966C153258448 @default.
- W1788809966 hasConceptScore W1788809966C154945302 @default.
- W1788809966 hasConceptScore W1788809966C158968445 @default.
- W1788809966 hasConceptScore W1788809966C196921405 @default.
- W1788809966 hasConceptScore W1788809966C206688291 @default.
- W1788809966 hasConceptScore W1788809966C2776135515 @default.
- W1788809966 hasConceptScore W1788809966C2776637919 @default.
- W1788809966 hasConceptScore W1788809966C2780069185 @default.
- W1788809966 hasConceptScore W1788809966C33923547 @default.
- W1788809966 hasConceptScore W1788809966C41008148 @default.
- W1788809966 hasConceptScore W1788809966C50644808 @default.
- W1788809966 hasLocation W17888099661 @default.
- W1788809966 hasOpenAccess W1788809966 @default.
- W1788809966 hasPrimaryLocation W17888099661 @default.
- W1788809966 hasRelatedWork W131378802 @default.
- W1788809966 hasRelatedWork W1505731132 @default.
- W1788809966 hasRelatedWork W1518461880 @default.
- W1788809966 hasRelatedWork W1570963478 @default.
- W1788809966 hasRelatedWork W1988790447 @default.
- W1788809966 hasRelatedWork W2016384870 @default.
- W1788809966 hasRelatedWork W2074694452 @default.
- W1788809966 hasRelatedWork W2077723394 @default.
- W1788809966 hasRelatedWork W2096199223 @default.
- W1788809966 hasRelatedWork W2135046866 @default.
- W1788809966 hasRelatedWork W2139759436 @default.
- W1788809966 hasRelatedWork W2146502635 @default.
- W1788809966 hasRelatedWork W2148825261 @default.
- W1788809966 hasRelatedWork W2164301055 @default.
- W1788809966 hasRelatedWork W2169713291 @default.
- W1788809966 hasRelatedWork W2296319761 @default.
- W1788809966 hasRelatedWork W2475334473 @default.
- W1788809966 hasRelatedWork W2513180554 @default.
- W1788809966 hasRelatedWork W2616657226 @default.
- W1788809966 hasRelatedWork W2964323557 @default.
- W1788809966 isParatext "false" @default.
- W1788809966 isRetracted "false" @default.
- W1788809966 magId "1788809966" @default.
- W1788809966 workType "article" @default.