Matches in SemOpenAlex for { <https://semopenalex.org/work/W3099934927> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W3099934927 abstract "Machine learning pipelines often rely on optimization procedures to make discrete decisions (e.g. sorting, picking closest neighbors, finding shortest paths or optimal matchings). Although these discrete decisions are easily computed in a forward manner, they cannot be used to modify model parameters using first-order optimization techniques because they break the back-propagation of computational graphs. In order to expand the scope of learning problems that can be solved in an end-to-end fashion, we propose a systematic method to transform a block that outputs an optimal discrete decision into a differentiable operation. Our approach relies on stochastic perturbations of these parameters, and can be used readily within existing solvers without the need for ad hoc regularization or smoothing. These perturbed optimizers yield solutions that are differentiable and never locally constant. The amount of smoothness can be tuned via the chosen noise amplitude, whose impact we analyze. The derivatives of these perturbed solvers can be evaluated eciently. We also show how this framework can be connected to a family of losses developed in structured prediction, and describe how these can be used in unsupervised and supervised learning, with theoretical guarantees. We demonstrate the performance of our approach on several machine learning tasks in experiments on synthetic and real data." @default.
- W3099934927 created "2020-11-23" @default.
- W3099934927 creator A5001483226 @default.
- W3099934927 creator A5017263427 @default.
- W3099934927 creator A5024896990 @default.
- W3099934927 creator A5049123454 @default.
- W3099934927 creator A5050982379 @default.
- W3099934927 creator A5064770739 @default.
- W3099934927 date "2020-02-20" @default.
- W3099934927 modified "2023-09-26" @default.
- W3099934927 title "Learning with differentiable perturbed optimizers" @default.
- W3099934927 hasPublicationYear "2020" @default.
- W3099934927 type Work @default.
- W3099934927 sameAs 3099934927 @default.
- W3099934927 citedByCount "9" @default.
- W3099934927 countsByYear W30999349272021 @default.
- W3099934927 countsByYear W30999349272022 @default.
- W3099934927 crossrefType "journal-article" @default.
- W3099934927 hasAuthorship W3099934927A5001483226 @default.
- W3099934927 hasAuthorship W3099934927A5017263427 @default.
- W3099934927 hasAuthorship W3099934927A5024896990 @default.
- W3099934927 hasAuthorship W3099934927A5049123454 @default.
- W3099934927 hasAuthorship W3099934927A5050982379 @default.
- W3099934927 hasAuthorship W3099934927A5064770739 @default.
- W3099934927 hasConcept C102634674 @default.
- W3099934927 hasConcept C11413529 @default.
- W3099934927 hasConcept C126255220 @default.
- W3099934927 hasConcept C134306372 @default.
- W3099934927 hasConcept C154945302 @default.
- W3099934927 hasConcept C202615002 @default.
- W3099934927 hasConcept C2776135515 @default.
- W3099934927 hasConcept C31972630 @default.
- W3099934927 hasConcept C33923547 @default.
- W3099934927 hasConcept C3770464 @default.
- W3099934927 hasConcept C41008148 @default.
- W3099934927 hasConceptScore W3099934927C102634674 @default.
- W3099934927 hasConceptScore W3099934927C11413529 @default.
- W3099934927 hasConceptScore W3099934927C126255220 @default.
- W3099934927 hasConceptScore W3099934927C134306372 @default.
- W3099934927 hasConceptScore W3099934927C154945302 @default.
- W3099934927 hasConceptScore W3099934927C202615002 @default.
- W3099934927 hasConceptScore W3099934927C2776135515 @default.
- W3099934927 hasConceptScore W3099934927C31972630 @default.
- W3099934927 hasConceptScore W3099934927C33923547 @default.
- W3099934927 hasConceptScore W3099934927C3770464 @default.
- W3099934927 hasConceptScore W3099934927C41008148 @default.
- W3099934927 hasLocation W30999349271 @default.
- W3099934927 hasOpenAccess W3099934927 @default.
- W3099934927 hasPrimaryLocation W30999349271 @default.
- W3099934927 hasRelatedWork W1483864446 @default.
- W3099934927 hasRelatedWork W1536680647 @default.
- W3099934927 hasRelatedWork W1537765963 @default.
- W3099934927 hasRelatedWork W1547176662 @default.
- W3099934927 hasRelatedWork W1606526398 @default.
- W3099934927 hasRelatedWork W1933411530 @default.
- W3099934927 hasRelatedWork W1984608801 @default.
- W3099934927 hasRelatedWork W2194775991 @default.
- W3099934927 hasRelatedWork W2247448034 @default.
- W3099934927 hasRelatedWork W2963970238 @default.
- W3099934927 hasRelatedWork W2964121744 @default.
- W3099934927 hasRelatedWork W2964218708 @default.
- W3099934927 hasRelatedWork W2971011990 @default.
- W3099934927 hasRelatedWork W2991650858 @default.
- W3099934927 hasRelatedWork W2995194170 @default.
- W3099934927 hasRelatedWork W3040996572 @default.
- W3099934927 hasRelatedWork W3044725263 @default.
- W3099934927 hasRelatedWork W3080700717 @default.
- W3099934927 hasRelatedWork W3098261982 @default.
- W3099934927 hasRelatedWork W3103412713 @default.
- W3099934927 isParatext "false" @default.
- W3099934927 isRetracted "false" @default.
- W3099934927 magId "3099934927" @default.
- W3099934927 workType "article" @default.