Matches in SemOpenAlex for { <https://semopenalex.org/work/W4379115893> ?p ?o ?g. }
Showing items 1 to 54 of
54
with 100 items per page.
- W4379115893 abstract "Post-training quantization of neural networks consists in quantizing a model without retraining nor hyperparameter search, while being fast and data frugal. In this paper, we propose LatticeQ, a novel post-training weight quantization method designed for deep convolutional neural networks (DC-NNs). Contrary to scalar rounding widely used in state-of-the-art quantization methods, LatticeQ uses a quantizer based on lattices - discrete algebraic structures. LatticeQ exploits the inner correlations between the model parameters to the benefit of minimizing quantization error. We achieve state-of-the-art results in post-training quantization. In particular, we achieve ImageNet classification results close to full precision on Resnet-18/50, with little to no accuracy drop for 4-bit models. Our code is available here, and a more thorough version of the paper here." @default.
- W4379115893 created "2023-06-03" @default.
- W4379115893 creator A5008973296 @default.
- W4379115893 creator A5009226493 @default.
- W4379115893 creator A5023918245 @default.
- W4379115893 creator A5052581839 @default.
- W4379115893 creator A5076632990 @default.
- W4379115893 date "2023-04-01" @default.
- W4379115893 modified "2023-09-30" @default.
- W4379115893 title "Lattice Quantization" @default.
- W4379115893 doi "https://doi.org/10.23919/date56975.2023.10137188" @default.
- W4379115893 hasPublicationYear "2023" @default.
- W4379115893 type Work @default.
- W4379115893 citedByCount "0" @default.
- W4379115893 crossrefType "proceedings-article" @default.
- W4379115893 hasAuthorship W4379115893A5008973296 @default.
- W4379115893 hasAuthorship W4379115893A5009226493 @default.
- W4379115893 hasAuthorship W4379115893A5023918245 @default.
- W4379115893 hasAuthorship W4379115893A5052581839 @default.
- W4379115893 hasAuthorship W4379115893A5076632990 @default.
- W4379115893 hasConcept C111919701 @default.
- W4379115893 hasConcept C11413529 @default.
- W4379115893 hasConcept C136625980 @default.
- W4379115893 hasConcept C154945302 @default.
- W4379115893 hasConcept C28855332 @default.
- W4379115893 hasConcept C41008148 @default.
- W4379115893 hasConcept C50644808 @default.
- W4379115893 hasConcept C81363708 @default.
- W4379115893 hasConcept C8642999 @default.
- W4379115893 hasConceptScore W4379115893C111919701 @default.
- W4379115893 hasConceptScore W4379115893C11413529 @default.
- W4379115893 hasConceptScore W4379115893C136625980 @default.
- W4379115893 hasConceptScore W4379115893C154945302 @default.
- W4379115893 hasConceptScore W4379115893C28855332 @default.
- W4379115893 hasConceptScore W4379115893C41008148 @default.
- W4379115893 hasConceptScore W4379115893C50644808 @default.
- W4379115893 hasConceptScore W4379115893C81363708 @default.
- W4379115893 hasConceptScore W4379115893C8642999 @default.
- W4379115893 hasLocation W43791158931 @default.
- W4379115893 hasOpenAccess W4379115893 @default.
- W4379115893 hasPrimaryLocation W43791158931 @default.
- W4379115893 hasRelatedWork W2141427150 @default.
- W4379115893 hasRelatedWork W2964106404 @default.
- W4379115893 hasRelatedWork W3045215719 @default.
- W4379115893 hasRelatedWork W3130227562 @default.
- W4379115893 hasRelatedWork W3132289026 @default.
- W4379115893 hasRelatedWork W4206951940 @default.
- W4379115893 hasRelatedWork W4289438529 @default.
- W4379115893 hasRelatedWork W4293868382 @default.
- W4379115893 hasRelatedWork W4304182771 @default.
- W4379115893 hasRelatedWork W4379115893 @default.
- W4379115893 isParatext "false" @default.
- W4379115893 isRetracted "false" @default.
- W4379115893 workType "article" @default.