Matches in SemOpenAlex for { <https://semopenalex.org/work/W2803858435> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W2803858435 abstract "The training phases of Deep neural network~(DNN) consumes enormous processing time and energy. Compression techniques utilizing the sparsity of DNNs can effectively accelerate the inference phase of DNNs. However, it can be hardly used in the training phase because the training phase involves dense matrix-multiplication using General Purpose Computation on Graphics Processors (GPGPU), which endorse regular and structural data layout. In this paper, we propose the Approximate Random Dropout that replaces the conventional random dropout of neurons and synapses with a regular and predefined patterns to eliminate the unnecessary computation and data access. To compensate the potential performance loss we develop a SGD-based Search Algorithm to produce the distribution of dropout patterns. We prove our approach is statistically equivalent to the previous dropout method. Experiments results on MLP and LSTM using well-known benchmarks show that the proposed Approximate Random Dropout can reduce the training time by $20%$-$77%$ ($19%$-$60%$) when dropout rate is $0.3$-$0.7$ on MLP (LSTM) with marginal accuracy drop." @default.
- W2803858435 created "2018-06-01" @default.
- W2803858435 creator A5011544360 @default.
- W2803858435 creator A5053801300 @default.
- W2803858435 creator A5056632010 @default.
- W2803858435 creator A5066720399 @default.
- W2803858435 creator A5066805530 @default.
- W2803858435 creator A5067161373 @default.
- W2803858435 creator A5070778210 @default.
- W2803858435 creator A5087474202 @default.
- W2803858435 date "2018-05-22" @default.
- W2803858435 modified "2023-10-18" @default.
- W2803858435 title "Approximate Random Dropout" @default.
- W2803858435 cites W148353443 @default.
- W2803858435 cites W1632114991 @default.
- W2803858435 cites W1724438581 @default.
- W2803858435 cites W2095705004 @default.
- W2803858435 cites W2168231600 @default.
- W2803858435 cites W2255466643 @default.
- W2803858435 cites W2271840356 @default.
- W2803858435 cites W2284050935 @default.
- W2803858435 cites W2513419314 @default.
- W2803858435 cites W2531409750 @default.
- W2803858435 cites W2556967412 @default.
- W2803858435 cites W2593245696 @default.
- W2803858435 cites W2612445135 @default.
- W2803858435 cites W2617766261 @default.
- W2803858435 cites W2622263826 @default.
- W2803858435 cites W2739789140 @default.
- W2803858435 cites W2742947407 @default.
- W2803858435 cites W2754526845 @default.
- W2803858435 cites W2785457004 @default.
- W2803858435 cites W2785896739 @default.
- W2803858435 cites W2912934387 @default.
- W2803858435 cites W2950094539 @default.
- W2803858435 cites W2950967261 @default.
- W2803858435 cites W2962786581 @default.
- W2803858435 cites W2962950660 @default.
- W2803858435 cites W2963125010 @default.
- W2803858435 cites W2964299589 @default.
- W2803858435 cites W2964325005 @default.
- W2803858435 cites W3104393472 @default.
- W2803858435 cites W3118608800 @default.
- W2803858435 cites W4919037 @default.
- W2803858435 doi "https://doi.org/10.48550/arxiv.1805.08939" @default.
- W2803858435 hasPublicationYear "2018" @default.
- W2803858435 type Work @default.
- W2803858435 sameAs 2803858435 @default.
- W2803858435 citedByCount "0" @default.
- W2803858435 crossrefType "posted-content" @default.
- W2803858435 hasAuthorship W2803858435A5011544360 @default.
- W2803858435 hasAuthorship W2803858435A5053801300 @default.
- W2803858435 hasAuthorship W2803858435A5056632010 @default.
- W2803858435 hasAuthorship W2803858435A5066720399 @default.
- W2803858435 hasAuthorship W2803858435A5066805530 @default.
- W2803858435 hasAuthorship W2803858435A5067161373 @default.
- W2803858435 hasAuthorship W2803858435A5070778210 @default.
- W2803858435 hasAuthorship W2803858435A5087474202 @default.
- W2803858435 hasBestOaLocation W28038584351 @default.
- W2803858435 hasConcept C11413529 @default.
- W2803858435 hasConcept C119857082 @default.
- W2803858435 hasConcept C154945302 @default.
- W2803858435 hasConcept C2776145597 @default.
- W2803858435 hasConcept C2776214188 @default.
- W2803858435 hasConcept C41008148 @default.
- W2803858435 hasConcept C45374587 @default.
- W2803858435 hasConcept C50644808 @default.
- W2803858435 hasConceptScore W2803858435C11413529 @default.
- W2803858435 hasConceptScore W2803858435C119857082 @default.
- W2803858435 hasConceptScore W2803858435C154945302 @default.
- W2803858435 hasConceptScore W2803858435C2776145597 @default.
- W2803858435 hasConceptScore W2803858435C2776214188 @default.
- W2803858435 hasConceptScore W2803858435C41008148 @default.
- W2803858435 hasConceptScore W2803858435C45374587 @default.
- W2803858435 hasConceptScore W2803858435C50644808 @default.
- W2803858435 hasLocation W28038584351 @default.
- W2803858435 hasOpenAccess W2803858435 @default.
- W2803858435 hasPrimaryLocation W28038584351 @default.
- W2803858435 hasRelatedWork W2211820962 @default.
- W2803858435 hasRelatedWork W2354062721 @default.
- W2803858435 hasRelatedWork W2386387936 @default.
- W2803858435 hasRelatedWork W2392110728 @default.
- W2803858435 hasRelatedWork W2582420655 @default.
- W2803858435 hasRelatedWork W2748952813 @default.
- W2803858435 hasRelatedWork W3107474891 @default.
- W2803858435 hasRelatedWork W3188973562 @default.
- W2803858435 hasRelatedWork W4205945679 @default.
- W2803858435 hasRelatedWork W4287244802 @default.
- W2803858435 isParatext "false" @default.
- W2803858435 isRetracted "false" @default.
- W2803858435 magId "2803858435" @default.
- W2803858435 workType "article" @default.