Matches in SemOpenAlex for { <https://semopenalex.org/work/W3038052560> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W3038052560 abstract "Abstract Brains process information in spiking neural networks. Their intricate connections shape the diverse functions these networks perform. In comparison, the functional capabilities of models of spiking networks are still rudimentary. This shortcoming is mainly due to the lack of insight and practical algorithms to construct the necessary connectivity. Any such algorithm typically attempts to build networks by iteratively reducing the error compared to a desired output. But assigning credit to hidden units in multi-layered spiking networks has remained challenging due to the non-differentiable nonlinearity of spikes. To avoid this issue, one can employ surrogate gradients to discover the required connectivity in spiking network models. However, the choice of a surrogate is not unique, raising the question of how its implementation influences the effectiveness of the method. Here, we use numerical simulations to systematically study how essential design parameters of surrogate gradients impact learning performance on a range of classification problems. We show that surrogate gradient learning is robust to different shapes of underlying surrogate derivatives, but the choice of the derivative’s scale can substantially affect learning performance. When we combine surrogate gradients with a suitable activity regularization technique, robust information processing can be achieved in spiking networks even at the sparse activity limit. Our study provides a systematic account of the remarkable robustness of surrogate gradient learning and serves as a practical guide to model functional spiking neural networks." @default.
- W3038052560 created "2020-07-02" @default.
- W3038052560 creator A5001875800 @default.
- W3038052560 creator A5085417824 @default.
- W3038052560 date "2020-06-29" @default.
- W3038052560 modified "2023-10-14" @default.
- W3038052560 title "The remarkable robustness of surrogate gradient learning for instilling complex function in spiking neural networks" @default.
- W3038052560 cites W1484004770 @default.
- W3038052560 cites W1677182931 @default.
- W3038052560 cites W1973327463 @default.
- W3038052560 cites W2003357516 @default.
- W3038052560 cites W2016589492 @default.
- W3038052560 cites W2047125104 @default.
- W3038052560 cites W2058616551 @default.
- W3038052560 cites W2069143585 @default.
- W3038052560 cites W2069552454 @default.
- W3038052560 cites W2076063813 @default.
- W3038052560 cites W2115831804 @default.
- W3038052560 cites W2274405424 @default.
- W3038052560 cites W2290982066 @default.
- W3038052560 cites W2314470091 @default.
- W3038052560 cites W2513853720 @default.
- W3038052560 cites W2595593680 @default.
- W3038052560 cites W2745933219 @default.
- W3038052560 cites W2772691327 @default.
- W3038052560 cites W2809808428 @default.
- W3038052560 cites W2896513464 @default.
- W3038052560 cites W2898323475 @default.
- W3038052560 cites W2899217980 @default.
- W3038052560 cites W2903763127 @default.
- W3038052560 cites W2907500638 @default.
- W3038052560 cites W2919115771 @default.
- W3038052560 cites W2962804204 @default.
- W3038052560 cites W2963119595 @default.
- W3038052560 cites W2963335874 @default.
- W3038052560 cites W2964017885 @default.
- W3038052560 cites W2978368159 @default.
- W3038052560 cites W2984844508 @default.
- W3038052560 cites W2990793844 @default.
- W3038052560 cites W3035400263 @default.
- W3038052560 cites W3098780744 @default.
- W3038052560 cites W3102087395 @default.
- W3038052560 cites W3102777511 @default.
- W3038052560 cites W4231081240 @default.
- W3038052560 doi "https://doi.org/10.1101/2020.06.29.176925" @default.
- W3038052560 hasPublicationYear "2020" @default.
- W3038052560 type Work @default.
- W3038052560 sameAs 3038052560 @default.
- W3038052560 citedByCount "8" @default.
- W3038052560 countsByYear W30380525602021 @default.
- W3038052560 countsByYear W30380525602022 @default.
- W3038052560 crossrefType "posted-content" @default.
- W3038052560 hasAuthorship W3038052560A5001875800 @default.
- W3038052560 hasAuthorship W3038052560A5085417824 @default.
- W3038052560 hasBestOaLocation W30380525601 @default.
- W3038052560 hasConcept C104317684 @default.
- W3038052560 hasConcept C11731999 @default.
- W3038052560 hasConcept C119857082 @default.
- W3038052560 hasConcept C131675550 @default.
- W3038052560 hasConcept C154945302 @default.
- W3038052560 hasConcept C185592680 @default.
- W3038052560 hasConcept C2776135515 @default.
- W3038052560 hasConcept C38365724 @default.
- W3038052560 hasConcept C41008148 @default.
- W3038052560 hasConcept C50644808 @default.
- W3038052560 hasConcept C55493867 @default.
- W3038052560 hasConcept C63479239 @default.
- W3038052560 hasConceptScore W3038052560C104317684 @default.
- W3038052560 hasConceptScore W3038052560C11731999 @default.
- W3038052560 hasConceptScore W3038052560C119857082 @default.
- W3038052560 hasConceptScore W3038052560C131675550 @default.
- W3038052560 hasConceptScore W3038052560C154945302 @default.
- W3038052560 hasConceptScore W3038052560C185592680 @default.
- W3038052560 hasConceptScore W3038052560C2776135515 @default.
- W3038052560 hasConceptScore W3038052560C38365724 @default.
- W3038052560 hasConceptScore W3038052560C41008148 @default.
- W3038052560 hasConceptScore W3038052560C50644808 @default.
- W3038052560 hasConceptScore W3038052560C55493867 @default.
- W3038052560 hasConceptScore W3038052560C63479239 @default.
- W3038052560 hasLocation W30380525601 @default.
- W3038052560 hasLocation W30380525602 @default.
- W3038052560 hasOpenAccess W3038052560 @default.
- W3038052560 hasPrimaryLocation W30380525601 @default.
- W3038052560 hasRelatedWork W2473399633 @default.
- W3038052560 hasRelatedWork W2910816794 @default.
- W3038052560 hasRelatedWork W2911843975 @default.
- W3038052560 hasRelatedWork W3110577345 @default.
- W3038052560 hasRelatedWork W3134817226 @default.
- W3038052560 hasRelatedWork W3198895633 @default.
- W3038052560 hasRelatedWork W4287019248 @default.
- W3038052560 hasRelatedWork W4318066384 @default.
- W3038052560 hasRelatedWork W4385451479 @default.
- W3038052560 hasRelatedWork W2946235959 @default.
- W3038052560 isParatext "false" @default.
- W3038052560 isRetracted "false" @default.
- W3038052560 magId "3038052560" @default.
- W3038052560 workType "article" @default.