Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386857650> ?p ?o ?g. }
- W4386857650 endingPage "106125" @default.
- W4386857650 startingPage "106103" @default.
- W4386857650 abstract "Matrix operations are nowadays central in many Machine Learning techniques, including in particular Deep Neural Networks (DNNs), whose core of any inference is represented by a sequence of dot product operations. An increasingly emerging problem is how to efficiently engineer their storage and operations. In this article we propose two new lossless compression schemes for real-valued matrices, supporting efficient vector-matrix multiplications in the compressed format, and specifically suitable for DNNs compression. Exploiting several recent studies that use weight pruning and quantization techniques to reduce the complexity of DNN inference, our schemes are expressly designed to benefit from both, that is from input matrices characterized by low entropy. In particular, our solutions are able to take advantage from the depth of the model, and the deeper the model, the higher the efficiency. Moreover, we derived space upper bounds for both variants in terms of the source entropy. Experiments show that our tools favourably compare in terms of energy and space efficiency against state-of-the-art matrix compression approaches, including Compressed Linear Algebra (CLA) and Compressed Shared Elements Row (CSER), the latter explicitly proposed in the context of DNN compression." @default.
- W4386857650 created "2023-09-20" @default.
- W4386857650 creator A5005837237 @default.
- W4386857650 creator A5026141403 @default.
- W4386857650 creator A5075385368 @default.
- W4386857650 creator A5091386422 @default.
- W4386857650 date "2023-01-01" @default.
- W4386857650 modified "2023-10-18" @default.
- W4386857650 title "Efficient and Compact Representations of Deep Neural Networks via Entropy Coding" @default.
- W4386857650 cites W1506342804 @default.
- W4386857650 cites W1524097605 @default.
- W4386857650 cites W1577505202 @default.
- W4386857650 cites W1944940353 @default.
- W4386857650 cites W1984541135 @default.
- W4386857650 cites W1988311079 @default.
- W4386857650 cites W1995875735 @default.
- W4386857650 cites W2014008637 @default.
- W4386857650 cites W2021061482 @default.
- W4386857650 cites W2023019000 @default.
- W4386857650 cites W2035080386 @default.
- W4386857650 cites W2047244756 @default.
- W4386857650 cites W2060108852 @default.
- W4386857650 cites W2070188654 @default.
- W4386857650 cites W2076789254 @default.
- W4386857650 cites W2086286404 @default.
- W4386857650 cites W2108189215 @default.
- W4386857650 cites W2112796928 @default.
- W4386857650 cites W2117539524 @default.
- W4386857650 cites W2119112357 @default.
- W4386857650 cites W2128676746 @default.
- W4386857650 cites W2131578006 @default.
- W4386857650 cites W2146519989 @default.
- W4386857650 cites W2257979135 @default.
- W4386857650 cites W2535724050 @default.
- W4386857650 cites W2561887839 @default.
- W4386857650 cites W2570841452 @default.
- W4386857650 cites W2785947426 @default.
- W4386857650 cites W2798170643 @default.
- W4386857650 cites W2883111419 @default.
- W4386857650 cites W2884085626 @default.
- W4386857650 cites W2917355316 @default.
- W4386857650 cites W2941733496 @default.
- W4386857650 cites W2964051635 @default.
- W4386857650 cites W3006631416 @default.
- W4386857650 cites W3008725350 @default.
- W4386857650 cites W3012561096 @default.
- W4386857650 cites W3091486761 @default.
- W4386857650 cites W3162937989 @default.
- W4386857650 cites W3163177193 @default.
- W4386857650 cites W3163895379 @default.
- W4386857650 cites W3176053424 @default.
- W4386857650 cites W3212348580 @default.
- W4386857650 cites W4294904134 @default.
- W4386857650 cites W4310147199 @default.
- W4386857650 cites W4319873770 @default.
- W4386857650 cites W77200240 @default.
- W4386857650 doi "https://doi.org/10.1109/access.2023.3317293" @default.
- W4386857650 hasPublicationYear "2023" @default.
- W4386857650 type Work @default.
- W4386857650 citedByCount "0" @default.
- W4386857650 crossrefType "journal-article" @default.
- W4386857650 hasAuthorship W4386857650A5005837237 @default.
- W4386857650 hasAuthorship W4386857650A5026141403 @default.
- W4386857650 hasAuthorship W4386857650A5075385368 @default.
- W4386857650 hasAuthorship W4386857650A5091386422 @default.
- W4386857650 hasBestOaLocation W43868576501 @default.
- W4386857650 hasConcept C106301342 @default.
- W4386857650 hasConcept C11413529 @default.
- W4386857650 hasConcept C121332964 @default.
- W4386857650 hasConcept C154945302 @default.
- W4386857650 hasConcept C17349429 @default.
- W4386857650 hasConcept C1769480 @default.
- W4386857650 hasConcept C2524010 @default.
- W4386857650 hasConcept C2776214188 @default.
- W4386857650 hasConcept C32900221 @default.
- W4386857650 hasConcept C33923547 @default.
- W4386857650 hasConcept C41008148 @default.
- W4386857650 hasConcept C50644808 @default.
- W4386857650 hasConcept C62520636 @default.
- W4386857650 hasConcept C78548338 @default.
- W4386857650 hasConcept C80444323 @default.
- W4386857650 hasConcept C81081738 @default.
- W4386857650 hasConcept C84114770 @default.
- W4386857650 hasConceptScore W4386857650C106301342 @default.
- W4386857650 hasConceptScore W4386857650C11413529 @default.
- W4386857650 hasConceptScore W4386857650C121332964 @default.
- W4386857650 hasConceptScore W4386857650C154945302 @default.
- W4386857650 hasConceptScore W4386857650C17349429 @default.
- W4386857650 hasConceptScore W4386857650C1769480 @default.
- W4386857650 hasConceptScore W4386857650C2524010 @default.
- W4386857650 hasConceptScore W4386857650C2776214188 @default.
- W4386857650 hasConceptScore W4386857650C32900221 @default.
- W4386857650 hasConceptScore W4386857650C33923547 @default.
- W4386857650 hasConceptScore W4386857650C41008148 @default.
- W4386857650 hasConceptScore W4386857650C50644808 @default.
- W4386857650 hasConceptScore W4386857650C62520636 @default.
- W4386857650 hasConceptScore W4386857650C78548338 @default.
- W4386857650 hasConceptScore W4386857650C80444323 @default.