Matches in SemOpenAlex for { <https://semopenalex.org/work/W4308233003> ?p ?o ?g. }
Showing items 1 to 70 of
70
with 100 items per page.
- W4308233003 abstract "The Lipschitz constant of the map between the input and output space represented by a neural network is a natural metric for assessing the robustness of the model. We present a new method to constrain the Lipschitz constant of dense deep learning models that can also be generalized to other architectures. The method relies on a simple weight normalization scheme during training that ensures the Lipschitz constant of every layer is below an upper limit specified by the analyst. A simple monotonic residual connection can then be used to make the model monotonic in any subset of its inputs, which is useful in scenarios where domain knowledge dictates such dependence. Examples can be found in algorithmic fairness requirements or, as presented here, in the classification of the decays of subatomic particles produced at the CERN Large Hadron Collider. Our normalization is minimally constraining and allows the underlying architecture to maintain higher expressiveness compared to other techniques which aim to either control the Lipschitz constant of the model or ensure its monotonicity. We show how the algorithm was used to train a powerful, robust, and interpretable discriminator for heavy-flavor-quark decays, which has been adopted for use as the primary data-selection algorithm in the LHCb real-time data-processing system in the current LHC data-taking period known as Run 3. In addition, our algorithm has also achieved state-of-the-art performance on benchmarks in medicine, finance, and other applications." @default.
- W4308233003 created "2022-11-09" @default.
- W4308233003 creator A5038599717 @default.
- W4308233003 creator A5053579716 @default.
- W4308233003 creator A5068216895 @default.
- W4308233003 date "2021-11-30" @default.
- W4308233003 modified "2023-10-17" @default.
- W4308233003 title "Robust and Provably Monotonic Networks" @default.
- W4308233003 doi "https://doi.org/10.1088/2632-2153/aced80" @default.
- W4308233003 hasPublicationYear "2021" @default.
- W4308233003 type Work @default.
- W4308233003 citedByCount "0" @default.
- W4308233003 crossrefType "posted-content" @default.
- W4308233003 hasAuthorship W4308233003A5038599717 @default.
- W4308233003 hasAuthorship W4308233003A5053579716 @default.
- W4308233003 hasAuthorship W4308233003A5068216895 @default.
- W4308233003 hasBestOaLocation W43082330031 @default.
- W4308233003 hasConcept C104317684 @default.
- W4308233003 hasConcept C109214941 @default.
- W4308233003 hasConcept C11413529 @default.
- W4308233003 hasConcept C121332964 @default.
- W4308233003 hasConcept C134306372 @default.
- W4308233003 hasConcept C136886441 @default.
- W4308233003 hasConcept C144024400 @default.
- W4308233003 hasConcept C155512373 @default.
- W4308233003 hasConcept C185592680 @default.
- W4308233003 hasConcept C19165224 @default.
- W4308233003 hasConcept C202444582 @default.
- W4308233003 hasConcept C22324862 @default.
- W4308233003 hasConcept C33923547 @default.
- W4308233003 hasConcept C41008148 @default.
- W4308233003 hasConcept C55493867 @default.
- W4308233003 hasConcept C63479239 @default.
- W4308233003 hasConcept C72169020 @default.
- W4308233003 hasConcept C87668248 @default.
- W4308233003 hasConceptScore W4308233003C104317684 @default.
- W4308233003 hasConceptScore W4308233003C109214941 @default.
- W4308233003 hasConceptScore W4308233003C11413529 @default.
- W4308233003 hasConceptScore W4308233003C121332964 @default.
- W4308233003 hasConceptScore W4308233003C134306372 @default.
- W4308233003 hasConceptScore W4308233003C136886441 @default.
- W4308233003 hasConceptScore W4308233003C144024400 @default.
- W4308233003 hasConceptScore W4308233003C155512373 @default.
- W4308233003 hasConceptScore W4308233003C185592680 @default.
- W4308233003 hasConceptScore W4308233003C19165224 @default.
- W4308233003 hasConceptScore W4308233003C202444582 @default.
- W4308233003 hasConceptScore W4308233003C22324862 @default.
- W4308233003 hasConceptScore W4308233003C33923547 @default.
- W4308233003 hasConceptScore W4308233003C41008148 @default.
- W4308233003 hasConceptScore W4308233003C55493867 @default.
- W4308233003 hasConceptScore W4308233003C63479239 @default.
- W4308233003 hasConceptScore W4308233003C72169020 @default.
- W4308233003 hasConceptScore W4308233003C87668248 @default.
- W4308233003 hasLocation W43082330031 @default.
- W4308233003 hasLocation W43082330032 @default.
- W4308233003 hasOpenAccess W4308233003 @default.
- W4308233003 hasPrimaryLocation W43082330031 @default.
- W4308233003 hasRelatedWork W175486093 @default.
- W4308233003 hasRelatedWork W2397777611 @default.
- W4308233003 hasRelatedWork W2748952813 @default.
- W4308233003 hasRelatedWork W2897842840 @default.
- W4308233003 hasRelatedWork W2911623553 @default.
- W4308233003 hasRelatedWork W3035794324 @default.
- W4308233003 hasRelatedWork W3185235544 @default.
- W4308233003 hasRelatedWork W4297791327 @default.
- W4308233003 hasRelatedWork W4297807321 @default.
- W4308233003 hasRelatedWork W4380682190 @default.
- W4308233003 isParatext "false" @default.
- W4308233003 isRetracted "false" @default.
- W4308233003 workType "article" @default.