Matches in SemOpenAlex for { <https://semopenalex.org/work/W2890904175> ?p ?o ?g. }
Showing items 1 to 86 of
86
with 100 items per page.
- W2890904175 abstract "Converting an n-dimensional vector to a probability distribution over n objects is a commonly used component in many machine learning tasks like multiclass classification, multilabel classification, attention mechanisms etc. For this, several probability mapping functions have been proposed and employed in literature such as softmax, sum-normalization, spherical softmax, and sparsemax, but there is very little understanding in terms how they relate with each other. Further, none of the above formulations offer an explicit control over the degree of sparsity. To address this, we develop a unified framework that encompasses all these formulations as special cases. This framework ensures simple closed-form solutions and existence of sub-gradients suitable for learning via backpropagation. Within this framework, we propose two novel sparse formulations, sparsegen-lin and sparsehourglass, that seek to provide a control over the degree of desired sparsity. We further develop novel convex loss functions that help induce the behavior of aforementioned formulations in the multilabel classification setting, showing improved performance. We also demonstrate empirically that the proposed formulations, when used to compute attention weights, achieve better or comparable performance on standard seq2seq tasks like neural machine translation and abstractive summarization." @default.
- W2890904175 created "2018-09-27" @default.
- W2890904175 creator A5022764301 @default.
- W2890904175 creator A5038482152 @default.
- W2890904175 creator A5050036814 @default.
- W2890904175 creator A5058098723 @default.
- W2890904175 creator A5060561123 @default.
- W2890904175 creator A5086266146 @default.
- W2890904175 date "2018-10-29" @default.
- W2890904175 modified "2023-09-27" @default.
- W2890904175 title "On Controllable Sparse Alternatives to Softmax" @default.
- W2890904175 cites W1515851193 @default.
- W2890904175 cites W1843891098 @default.
- W2890904175 cites W1978259121 @default.
- W2890904175 cites W2574872930 @default.
- W2890904175 cites W2606101940 @default.
- W2890904175 cites W2951008357 @default.
- W2890904175 cites W2963123301 @default.
- W2890904175 cites W2963603213 @default.
- W2890904175 cites W3099884890 @default.
- W2890904175 hasPublicationYear "2018" @default.
- W2890904175 type Work @default.
- W2890904175 sameAs 2890904175 @default.
- W2890904175 citedByCount "0" @default.
- W2890904175 crossrefType "posted-content" @default.
- W2890904175 hasAuthorship W2890904175A5022764301 @default.
- W2890904175 hasAuthorship W2890904175A5038482152 @default.
- W2890904175 hasAuthorship W2890904175A5050036814 @default.
- W2890904175 hasAuthorship W2890904175A5058098723 @default.
- W2890904175 hasAuthorship W2890904175A5060561123 @default.
- W2890904175 hasAuthorship W2890904175A5086266146 @default.
- W2890904175 hasConcept C111472728 @default.
- W2890904175 hasConcept C119857082 @default.
- W2890904175 hasConcept C136886441 @default.
- W2890904175 hasConcept C138885662 @default.
- W2890904175 hasConcept C144024400 @default.
- W2890904175 hasConcept C153180895 @default.
- W2890904175 hasConcept C154945302 @default.
- W2890904175 hasConcept C155032097 @default.
- W2890904175 hasConcept C170858558 @default.
- W2890904175 hasConcept C188441871 @default.
- W2890904175 hasConcept C19165224 @default.
- W2890904175 hasConcept C2780586882 @default.
- W2890904175 hasConcept C41008148 @default.
- W2890904175 hasConcept C50644808 @default.
- W2890904175 hasConceptScore W2890904175C111472728 @default.
- W2890904175 hasConceptScore W2890904175C119857082 @default.
- W2890904175 hasConceptScore W2890904175C136886441 @default.
- W2890904175 hasConceptScore W2890904175C138885662 @default.
- W2890904175 hasConceptScore W2890904175C144024400 @default.
- W2890904175 hasConceptScore W2890904175C153180895 @default.
- W2890904175 hasConceptScore W2890904175C154945302 @default.
- W2890904175 hasConceptScore W2890904175C155032097 @default.
- W2890904175 hasConceptScore W2890904175C170858558 @default.
- W2890904175 hasConceptScore W2890904175C188441871 @default.
- W2890904175 hasConceptScore W2890904175C19165224 @default.
- W2890904175 hasConceptScore W2890904175C2780586882 @default.
- W2890904175 hasConceptScore W2890904175C41008148 @default.
- W2890904175 hasConceptScore W2890904175C50644808 @default.
- W2890904175 hasLocation W28909041751 @default.
- W2890904175 hasOpenAccess W2890904175 @default.
- W2890904175 hasPrimaryLocation W28909041751 @default.
- W2890904175 hasRelatedWork W1884762066 @default.
- W2890904175 hasRelatedWork W2102871403 @default.
- W2890904175 hasRelatedWork W2162221686 @default.
- W2890904175 hasRelatedWork W2166180593 @default.
- W2890904175 hasRelatedWork W2182562048 @default.
- W2890904175 hasRelatedWork W2191710671 @default.
- W2890904175 hasRelatedWork W2194835488 @default.
- W2890904175 hasRelatedWork W2405778879 @default.
- W2890904175 hasRelatedWork W2460519968 @default.
- W2890904175 hasRelatedWork W2471962125 @default.
- W2890904175 hasRelatedWork W2557827285 @default.
- W2890904175 hasRelatedWork W2805880533 @default.
- W2890904175 hasRelatedWork W2949579938 @default.
- W2890904175 hasRelatedWork W2950515451 @default.
- W2890904175 hasRelatedWork W2951006247 @default.
- W2890904175 hasRelatedWork W2964231757 @default.
- W2890904175 hasRelatedWork W2964290105 @default.
- W2890904175 hasRelatedWork W2980725178 @default.
- W2890904175 hasRelatedWork W3005234653 @default.
- W2890904175 hasRelatedWork W3157050624 @default.
- W2890904175 isParatext "false" @default.
- W2890904175 isRetracted "false" @default.
- W2890904175 magId "2890904175" @default.
- W2890904175 workType "article" @default.