Matches in SemOpenAlex for { <https://semopenalex.org/work/W3033803554> ?p ?o ?g. }
- W3033803554 abstract "Computation using brain-inspired spiking neural networks (SNNs) with neuromorphic hardware may offer orders of magnitude higher energy efficiency compared to the current analog neural networks (ANNs). Unfortunately, training SNNs with the same number of layers as state of the art ANNs remains a challenge. To our knowledge the only method which is successful in this regard is supervised training of ANN and then converting it to SNN. In this work we directly train deep SNNs using backpropagation with surrogate gradient and find that due to implicitly recurrent nature of feed forward SNN's the exploding or vanishing gradient problem severely hinders their training. We show that this problem can be solved by tuning the surrogate gradient function. We also propose using batch normalization from ANN literature on input currents of SNN neurons. Using these improvements we show that is is possible to train SNN with ResNet50 architecture on CIFAR100 and Imagenette object recognition datasets. The trained SNN falls behind in accuracy compared to analogous ANN but requires several orders of magnitude less inference time steps (as low as 10) to reach good accuracy compared to SNNs obtained by conversion from ANN which require on the order of 1000 time steps." @default.
- W3033803554 created "2020-06-12" @default.
- W3033803554 creator A5014425014 @default.
- W3033803554 creator A5015965173 @default.
- W3033803554 creator A5021964474 @default.
- W3033803554 creator A5026181269 @default.
- W3033803554 date "2020-06-08" @default.
- W3033803554 modified "2023-09-27" @default.
- W3033803554 title "Training Deep Spiking Neural Networks." @default.
- W3033803554 cites W1484004770 @default.
- W3033803554 cites W1484977481 @default.
- W3033803554 cites W1645800954 @default.
- W3033803554 cites W1677182931 @default.
- W3033803554 cites W2120475512 @default.
- W3033803554 cites W2296761366 @default.
- W3033803554 cites W2314470091 @default.
- W3033803554 cites W2569813014 @default.
- W3033803554 cites W2775079417 @default.
- W3033803554 cites W2795247881 @default.
- W3033803554 cites W2891530223 @default.
- W3033803554 cites W2896097393 @default.
- W3033803554 cites W2908510526 @default.
- W3033803554 cites W2942803402 @default.
- W3033803554 cites W2949117887 @default.
- W3033803554 cites W2962949994 @default.
- W3033803554 cites W2963304263 @default.
- W3033803554 cites W2963760575 @default.
- W3033803554 cites W2963834742 @default.
- W3033803554 cites W2984844508 @default.
- W3033803554 cites W3007283957 @default.
- W3033803554 cites W3009092561 @default.
- W3033803554 cites W3023721287 @default.
- W3033803554 cites W3038702532 @default.
- W3033803554 cites W3102087395 @default.
- W3033803554 hasPublicationYear "2020" @default.
- W3033803554 type Work @default.
- W3033803554 sameAs 3033803554 @default.
- W3033803554 citedByCount "2" @default.
- W3033803554 countsByYear W30338035542021 @default.
- W3033803554 crossrefType "posted-content" @default.
- W3033803554 hasAuthorship W3033803554A5014425014 @default.
- W3033803554 hasAuthorship W3033803554A5015965173 @default.
- W3033803554 hasAuthorship W3033803554A5021964474 @default.
- W3033803554 hasAuthorship W3033803554A5026181269 @default.
- W3033803554 hasConcept C11413529 @default.
- W3033803554 hasConcept C11731999 @default.
- W3033803554 hasConcept C119857082 @default.
- W3033803554 hasConcept C136886441 @default.
- W3033803554 hasConcept C144024400 @default.
- W3033803554 hasConcept C151927369 @default.
- W3033803554 hasConcept C153180895 @default.
- W3033803554 hasConcept C154945302 @default.
- W3033803554 hasConcept C155032097 @default.
- W3033803554 hasConcept C19165224 @default.
- W3033803554 hasConcept C2776214188 @default.
- W3033803554 hasConcept C2984842247 @default.
- W3033803554 hasConcept C3832189 @default.
- W3033803554 hasConcept C41008148 @default.
- W3033803554 hasConcept C45374587 @default.
- W3033803554 hasConcept C50644808 @default.
- W3033803554 hasConceptScore W3033803554C11413529 @default.
- W3033803554 hasConceptScore W3033803554C11731999 @default.
- W3033803554 hasConceptScore W3033803554C119857082 @default.
- W3033803554 hasConceptScore W3033803554C136886441 @default.
- W3033803554 hasConceptScore W3033803554C144024400 @default.
- W3033803554 hasConceptScore W3033803554C151927369 @default.
- W3033803554 hasConceptScore W3033803554C153180895 @default.
- W3033803554 hasConceptScore W3033803554C154945302 @default.
- W3033803554 hasConceptScore W3033803554C155032097 @default.
- W3033803554 hasConceptScore W3033803554C19165224 @default.
- W3033803554 hasConceptScore W3033803554C2776214188 @default.
- W3033803554 hasConceptScore W3033803554C2984842247 @default.
- W3033803554 hasConceptScore W3033803554C3832189 @default.
- W3033803554 hasConceptScore W3033803554C41008148 @default.
- W3033803554 hasConceptScore W3033803554C45374587 @default.
- W3033803554 hasConceptScore W3033803554C50644808 @default.
- W3033803554 hasLocation W30338035541 @default.
- W3033803554 hasOpenAccess W3033803554 @default.
- W3033803554 hasPrimaryLocation W30338035541 @default.
- W3033803554 hasRelatedWork W2113579565 @default.
- W3033803554 hasRelatedWork W2114267009 @default.
- W3033803554 hasRelatedWork W2625915976 @default.
- W3033803554 hasRelatedWork W2903485762 @default.
- W3033803554 hasRelatedWork W2994470909 @default.
- W3033803554 hasRelatedWork W3024338890 @default.
- W3033803554 hasRelatedWork W3026086625 @default.
- W3033803554 hasRelatedWork W3048171402 @default.
- W3033803554 hasRelatedWork W3090954756 @default.
- W3033803554 hasRelatedWork W3106955873 @default.
- W3033803554 hasRelatedWork W3113651745 @default.
- W3033803554 hasRelatedWork W3153458041 @default.
- W3033803554 hasRelatedWork W3171047464 @default.
- W3033803554 hasRelatedWork W3175918389 @default.
- W3033803554 hasRelatedWork W3177348357 @default.
- W3033803554 hasRelatedWork W3193620589 @default.
- W3033803554 hasRelatedWork W3197853100 @default.
- W3033803554 hasRelatedWork W3203274274 @default.
- W3033803554 hasRelatedWork W3211954089 @default.
- W3033803554 hasRelatedWork W3212171552 @default.
- W3033803554 isParatext "false" @default.