Matches in SemOpenAlex for { <https://semopenalex.org/work/W2950455323> ?p ?o ?g. }
- W2950455323 abstract "Many state-of-the-art neural models for NLP are heavily parameterized and thus memory inefficient. This paper proposes a series of lightweight and memory efficient neural architectures for a potpourri of natural language processing (NLP) tasks. To this end, our models exploit computation using Quaternion algebra and hypercomplex spaces, enabling not only expressive inter-component interactions but also significantly ($75%$) reduced parameter size due to lesser degrees of freedom in the Hamilton product. We propose Quaternion variants of models, giving rise to new architectures such as the Quaternion attention Model and Quaternion Transformer. Extensive experiments on a battery of NLP tasks demonstrates the utility of proposed Quaternion-inspired models, enabling up to $75%$ reduction in parameter size without significant loss in performance." @default.
- W2950455323 created "2019-06-27" @default.
- W2950455323 creator A5001659855 @default.
- W2950455323 creator A5007322337 @default.
- W2950455323 creator A5026059624 @default.
- W2950455323 creator A5034826937 @default.
- W2950455323 creator A5046206794 @default.
- W2950455323 creator A5049841140 @default.
- W2950455323 creator A5057786383 @default.
- W2950455323 creator A5075728689 @default.
- W2950455323 date "2019-06-11" @default.
- W2950455323 modified "2023-10-18" @default.
- W2950455323 title "Lightweight and Efficient Neural Natural Language Processing with Quaternion Networks" @default.
- W2950455323 cites W1010415138 @default.
- W2950455323 cites W2113459411 @default.
- W2950455323 cites W2145544171 @default.
- W2950455323 cites W2162456950 @default.
- W2950455323 cites W2164019165 @default.
- W2950455323 cites W2250539671 @default.
- W2950455323 cites W2251818205 @default.
- W2950455323 cites W2251939518 @default.
- W2950455323 cites W2607892599 @default.
- W2950455323 cites W2608787653 @default.
- W2950455323 cites W2618946976 @default.
- W2950455323 cites W2741990823 @default.
- W2950455323 cites W2788496822 @default.
- W2950455323 cites W2808281579 @default.
- W2950455323 cites W2889747665 @default.
- W2950455323 cites W2893578842 @default.
- W2950455323 cites W2894977338 @default.
- W2950455323 cites W2905270607 @default.
- W2950455323 cites W2905556235 @default.
- W2950455323 cites W2942026896 @default.
- W2950455323 cites W2948937670 @default.
- W2950455323 cites W2950414499 @default.
- W2950455323 cites W2950738719 @default.
- W2950455323 cites W2951815760 @default.
- W2950455323 cites W2952113915 @default.
- W2950455323 cites W2952136670 @default.
- W2950455323 cites W2953084091 @default.
- W2950455323 cites W2953384591 @default.
- W2950455323 cites W2962916650 @default.
- W2950455323 cites W2963042606 @default.
- W2950455323 cites W2963230471 @default.
- W2950455323 cites W2963235055 @default.
- W2950455323 cites W2963341956 @default.
- W2950455323 cites W2963403868 @default.
- W2950455323 cites W2963432357 @default.
- W2950455323 cites W2963719423 @default.
- W2950455323 cites W2964308564 @default.
- W2950455323 cites W82661287 @default.
- W2950455323 cites W836999996 @default.
- W2950455323 doi "https://doi.org/10.48550/arxiv.1906.04393" @default.
- W2950455323 hasPublicationYear "2019" @default.
- W2950455323 type Work @default.
- W2950455323 sameAs 2950455323 @default.
- W2950455323 citedByCount "9" @default.
- W2950455323 countsByYear W29504553232019 @default.
- W2950455323 countsByYear W29504553232020 @default.
- W2950455323 countsByYear W29504553232021 @default.
- W2950455323 countsByYear W29504553232023 @default.
- W2950455323 crossrefType "posted-content" @default.
- W2950455323 hasAuthorship W2950455323A5001659855 @default.
- W2950455323 hasAuthorship W2950455323A5007322337 @default.
- W2950455323 hasAuthorship W2950455323A5026059624 @default.
- W2950455323 hasAuthorship W2950455323A5034826937 @default.
- W2950455323 hasAuthorship W2950455323A5046206794 @default.
- W2950455323 hasAuthorship W2950455323A5049841140 @default.
- W2950455323 hasAuthorship W2950455323A5057786383 @default.
- W2950455323 hasAuthorship W2950455323A5075728689 @default.
- W2950455323 hasBestOaLocation W29504553231 @default.
- W2950455323 hasConcept C100899422 @default.
- W2950455323 hasConcept C11413529 @default.
- W2950455323 hasConcept C136119220 @default.
- W2950455323 hasConcept C14394260 @default.
- W2950455323 hasConcept C151640129 @default.
- W2950455323 hasConcept C154945302 @default.
- W2950455323 hasConcept C165464430 @default.
- W2950455323 hasConcept C200127275 @default.
- W2950455323 hasConcept C202444582 @default.
- W2950455323 hasConcept C203249530 @default.
- W2950455323 hasConcept C2524010 @default.
- W2950455323 hasConcept C33923547 @default.
- W2950455323 hasConcept C41008148 @default.
- W2950455323 hasConcept C45374587 @default.
- W2950455323 hasConcept C50644808 @default.
- W2950455323 hasConcept C80444323 @default.
- W2950455323 hasConceptScore W2950455323C100899422 @default.
- W2950455323 hasConceptScore W2950455323C11413529 @default.
- W2950455323 hasConceptScore W2950455323C136119220 @default.
- W2950455323 hasConceptScore W2950455323C14394260 @default.
- W2950455323 hasConceptScore W2950455323C151640129 @default.
- W2950455323 hasConceptScore W2950455323C154945302 @default.
- W2950455323 hasConceptScore W2950455323C165464430 @default.
- W2950455323 hasConceptScore W2950455323C200127275 @default.
- W2950455323 hasConceptScore W2950455323C202444582 @default.
- W2950455323 hasConceptScore W2950455323C203249530 @default.
- W2950455323 hasConceptScore W2950455323C2524010 @default.
- W2950455323 hasConceptScore W2950455323C33923547 @default.
- W2950455323 hasConceptScore W2950455323C41008148 @default.