FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2952023274> ?p ?o ?g. }

• W2952023274 abstract "Modern applications in sensitive domains such as biometrics and medicine frequently require the use of non-decomposable loss functions such as precision@k, F-measure etc. Compared to point loss functions such as hinge-loss, these offer much more fine grained control over prediction, but at the same time present novel challenges in terms of algorithm design and analysis. In this work we initiate a study of online learning techniques for such non-decomposable loss functions with an aim to enable incremental learning as well as design scalable solvers for batch problems. To this end, we propose an online learning framework for such loss functions. Our model enjoys several nice properties, chief amongst them being the existence of efficient online learning algorithms with sublinear regret and online to batch conversion bounds. Our model is a provable extension of existing online learning models for point loss functions. We instantiate two popular losses, prec@k and pAUC, in our model and prove sublinear regret bounds for both of them. Our proofs require a novel structural lemma over ranked lists which may be of independent interest. We then develop scalable stochastic gradient descent solvers for non-decomposable loss functions. We show that for a large family of loss functions satisfying a certain uniform convergence property (that includes prec@k, pAUC, and F-measure), our methods provably converge to the empirical risk minimizer. Such uniform convergence results were not known for these losses and we establish these using novel proof techniques. We then use extensive experimentation on real life and benchmark datasets to establish that our method can be orders of magnitude faster than a recently proposed cutting plane method." @default.
• W2952023274 created "2019-06-27" @default.
• W2952023274 creator A5034432097 @default.
• W2952023274 creator A5077482746 @default.
• W2952023274 creator A5081725635 @default.
• W2952023274 date "2014-10-24" @default.
• W2952023274 modified "2023-09-24" @default.
• W2952023274 title "Online and Stochastic Gradient Methods for Non-decomposable Loss Functions" @default.
• W2952023274 cites W1551269584 @default.
• W2952023274 cites W1604179126 @default.
• W2952023274 cites W1978810924 @default.
• W2952023274 cites W2070771761 @default.
• W2952023274 cites W2127176025 @default.
• W2952023274 cites W2130062883 @default.
• W2952023274 cites W2621384249 @default.
• W2952023274 cites W2950403298 @default.
• W2952023274 hasPublicationYear "2014" @default.
• W2952023274 type Work @default.
• W2952023274 sameAs 2952023274 @default.
• W2952023274 citedByCount "19" @default.
• W2952023274 countsByYear W29520232742015 @default.
• W2952023274 countsByYear W29520232742016 @default.
• W2952023274 countsByYear W29520232742017 @default.
• W2952023274 countsByYear W29520232742018 @default.
• W2952023274 countsByYear W29520232742019 @default.
• W2952023274 countsByYear W29520232742020 @default.
• W2952023274 countsByYear W29520232742021 @default.
• W2952023274 crossrefType "posted-content" @default.
• W2952023274 hasAuthorship W2952023274A5034432097 @default.
• W2952023274 hasAuthorship W2952023274A5077482746 @default.
• W2952023274 hasAuthorship W2952023274A5081725635 @default.
• W2952023274 hasConcept C108710211 @default.
• W2952023274 hasConcept C11413529 @default.
• W2952023274 hasConcept C117160843 @default.
• W2952023274 hasConcept C118615104 @default.
• W2952023274 hasConcept C119857082 @default.
• W2952023274 hasConcept C12267149 @default.
• W2952023274 hasConcept C124101348 @default.
• W2952023274 hasConcept C126255220 @default.
• W2952023274 hasConcept C162324750 @default.
• W2952023274 hasConcept C18903297 @default.
• W2952023274 hasConcept C206688291 @default.
• W2952023274 hasConcept C2524010 @default.
• W2952023274 hasConcept C2777303404 @default.
• W2952023274 hasConcept C2777759810 @default.
• W2952023274 hasConcept C2780009758 @default.
• W2952023274 hasConcept C33923547 @default.
• W2952023274 hasConcept C39891107 @default.
• W2952023274 hasConcept C41008148 @default.
• W2952023274 hasConcept C46757340 @default.
• W2952023274 hasConcept C48044578 @default.
• W2952023274 hasConcept C50522688 @default.
• W2952023274 hasConcept C50644808 @default.
• W2952023274 hasConcept C77088390 @default.
• W2952023274 hasConcept C80444323 @default.
• W2952023274 hasConcept C86803240 @default.
• W2952023274 hasConceptScore W2952023274C108710211 @default.
• W2952023274 hasConceptScore W2952023274C11413529 @default.
• W2952023274 hasConceptScore W2952023274C117160843 @default.
• W2952023274 hasConceptScore W2952023274C118615104 @default.
• W2952023274 hasConceptScore W2952023274C119857082 @default.
• W2952023274 hasConceptScore W2952023274C12267149 @default.
• W2952023274 hasConceptScore W2952023274C124101348 @default.
• W2952023274 hasConceptScore W2952023274C126255220 @default.
• W2952023274 hasConceptScore W2952023274C162324750 @default.
• W2952023274 hasConceptScore W2952023274C18903297 @default.
• W2952023274 hasConceptScore W2952023274C206688291 @default.
• W2952023274 hasConceptScore W2952023274C2524010 @default.
• W2952023274 hasConceptScore W2952023274C2777303404 @default.
• W2952023274 hasConceptScore W2952023274C2777759810 @default.
• W2952023274 hasConceptScore W2952023274C2780009758 @default.
• W2952023274 hasConceptScore W2952023274C33923547 @default.
• W2952023274 hasConceptScore W2952023274C39891107 @default.
• W2952023274 hasConceptScore W2952023274C41008148 @default.
• W2952023274 hasConceptScore W2952023274C46757340 @default.
• W2952023274 hasConceptScore W2952023274C48044578 @default.
• W2952023274 hasConceptScore W2952023274C50522688 @default.
• W2952023274 hasConceptScore W2952023274C50644808 @default.
• W2952023274 hasConceptScore W2952023274C77088390 @default.
• W2952023274 hasConceptScore W2952023274C80444323 @default.
• W2952023274 hasConceptScore W2952023274C86803240 @default.
• W2952023274 hasLocation W29520232741 @default.
• W2952023274 hasOpenAccess W2952023274 @default.
• W2952023274 hasPrimaryLocation W29520232741 @default.
• W2952023274 hasRelatedWork W2031248101 @default.
• W2952023274 hasRelatedWork W2070771761 @default.
• W2952023274 hasRelatedWork W2104144433 @default.
• W2952023274 hasRelatedWork W2109689843 @default.
• W2952023274 hasRelatedWork W2122770142 @default.
• W2952023274 hasRelatedWork W2148825261 @default.
• W2952023274 hasRelatedWork W2149433471 @default.
• W2952023274 hasRelatedWork W2201177080 @default.
• W2952023274 hasRelatedWork W2399077675 @default.
• W2952023274 hasRelatedWork W2491144192 @default.
• W2952023274 hasRelatedWork W2803448454 @default.
• W2952023274 hasRelatedWork W2951054453 @default.
• W2952023274 hasRelatedWork W2952550984 @default.
• W2952023274 hasRelatedWork W2952758290 @default.
• W2952023274 hasRelatedWork W2964027161 @default.
• W2952023274 hasRelatedWork W2964069971 @default.

• next