Matches in SemOpenAlex for { <https://semopenalex.org/work/W2892446824> ?p ?o ?g. }
Showing items 1 to 71 of
71
with 100 items per page.
- W2892446824 abstract "Compression is a key step to deploy large neural networks on resource-constrained platforms. As a popular compression technique, quantization constrains the number of distinct weight values and thus reducing the number of bits required to represent and store each weight. In this paper, we study the representation power of quantized neural networks. First, we prove the universal approximability of quantized ReLU networks on a wide class of functions. Then we provide upper bounds on the number of weights and the memory size for a given approximation error bound and the bit-width of weights for function-independent and function-dependent structures. Our results reveal that, to attain an approximation error bound of $epsilon$, the number of weights needed by a quantized network is no more than $mathcal{O}left(log^5(1/epsilon)right)$ times that of an unquantized network. This overhead is of much lower order than the lower bound of the number of weights needed for the error bound, supporting the empirical success of various quantization techniques. To the best of our knowledge, this is the first in-depth study on the complexity bounds of quantized neural networks." @default.
- W2892446824 created "2018-10-05" @default.
- W2892446824 creator A5000141831 @default.
- W2892446824 creator A5030156276 @default.
- W2892446824 creator A5051737658 @default.
- W2892446824 creator A5056712638 @default.
- W2892446824 date "2018-02-10" @default.
- W2892446824 modified "2023-09-25" @default.
- W2892446824 title "On the Universal Approximability and Complexity Bounds of Quantized ReLU Neural Networks" @default.
- W2892446824 doi "https://doi.org/10.48550/arxiv.1802.03646" @default.
- W2892446824 hasPublicationYear "2018" @default.
- W2892446824 type Work @default.
- W2892446824 sameAs 2892446824 @default.
- W2892446824 citedByCount "10" @default.
- W2892446824 countsByYear W28924468242018 @default.
- W2892446824 countsByYear W28924468242019 @default.
- W2892446824 countsByYear W28924468242020 @default.
- W2892446824 countsByYear W28924468242021 @default.
- W2892446824 crossrefType "posted-content" @default.
- W2892446824 hasAuthorship W2892446824A5000141831 @default.
- W2892446824 hasAuthorship W2892446824A5030156276 @default.
- W2892446824 hasAuthorship W2892446824A5051737658 @default.
- W2892446824 hasAuthorship W2892446824A5056712638 @default.
- W2892446824 hasBestOaLocation W28924468241 @default.
- W2892446824 hasConcept C111919701 @default.
- W2892446824 hasConcept C11413529 @default.
- W2892446824 hasConcept C118615104 @default.
- W2892446824 hasConcept C122383733 @default.
- W2892446824 hasConcept C134306372 @default.
- W2892446824 hasConcept C14036430 @default.
- W2892446824 hasConcept C154945302 @default.
- W2892446824 hasConcept C2779960059 @default.
- W2892446824 hasConcept C28855332 @default.
- W2892446824 hasConcept C33923547 @default.
- W2892446824 hasConcept C41008148 @default.
- W2892446824 hasConcept C50644808 @default.
- W2892446824 hasConcept C77553402 @default.
- W2892446824 hasConcept C78458016 @default.
- W2892446824 hasConcept C86803240 @default.
- W2892446824 hasConceptScore W2892446824C111919701 @default.
- W2892446824 hasConceptScore W2892446824C11413529 @default.
- W2892446824 hasConceptScore W2892446824C118615104 @default.
- W2892446824 hasConceptScore W2892446824C122383733 @default.
- W2892446824 hasConceptScore W2892446824C134306372 @default.
- W2892446824 hasConceptScore W2892446824C14036430 @default.
- W2892446824 hasConceptScore W2892446824C154945302 @default.
- W2892446824 hasConceptScore W2892446824C2779960059 @default.
- W2892446824 hasConceptScore W2892446824C28855332 @default.
- W2892446824 hasConceptScore W2892446824C33923547 @default.
- W2892446824 hasConceptScore W2892446824C41008148 @default.
- W2892446824 hasConceptScore W2892446824C50644808 @default.
- W2892446824 hasConceptScore W2892446824C77553402 @default.
- W2892446824 hasConceptScore W2892446824C78458016 @default.
- W2892446824 hasConceptScore W2892446824C86803240 @default.
- W2892446824 hasLocation W28924468241 @default.
- W2892446824 hasOpenAccess W2892446824 @default.
- W2892446824 hasPrimaryLocation W28924468241 @default.
- W2892446824 hasRelatedWork W2219767525 @default.
- W2892446824 hasRelatedWork W2355070670 @default.
- W2892446824 hasRelatedWork W2892001478 @default.
- W2892446824 hasRelatedWork W2892446824 @default.
- W2892446824 hasRelatedWork W2953328899 @default.
- W2892446824 hasRelatedWork W2963487827 @default.
- W2892446824 hasRelatedWork W2966900502 @default.
- W2892446824 hasRelatedWork W3087818803 @default.
- W2892446824 hasRelatedWork W4289306510 @default.
- W2892446824 hasRelatedWork W4303856917 @default.
- W2892446824 isParatext "false" @default.
- W2892446824 isRetracted "false" @default.
- W2892446824 magId "2892446824" @default.
- W2892446824 workType "article" @default.