FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3133996370> ?p ?o ?g. }

• W3133996370 abstract "Spiking neural networks (SNNs) offer an inherent ability to process spatial-temporal data, or in other words, realworld sensory data, but suffer from the difficulty of training high models. A major thread of research on SNNs is on converting a pre-trained convolutional neural network (CNN) to an SNN of the same structure. State-of-the-art conversion methods are approaching the limit, i.e., the near-zero loss of SNN against the original CNN. However, we note that this is made possible only when significantly more is consumed to process an input. In this paper, we argue that this trend of energy for accuracy is not necessary -- a little can go a long way to achieve the near-zero loss. Specifically, we propose a novel CNN-to-SNN conversion method that is able to use a reasonably short spike train (e.g., 256 timesteps for CIFAR10 images) to achieve the near-zero loss. The new conversion method, named as explicit current control (ECC), contains three techniques (current normalisation, thresholding for residual elimination, and consistency maintenance for batch-normalisation), in order to explicitly control the currents flowing through the SNN when processing inputs. We implement ECC into a tool nicknamed SpKeras, which can conveniently import Keras CNN models and convert them into SNNs. We conduct an extensive set of experiments with the tool -- working with VGG16 and various datasets such as CIFAR10 and CIFAR100 -- and compare with state-of-the-art conversion methods. Results show that ECC is a promising method that can optimise over consumption and loss simultaneously." @default.
• W3133996370 created "2021-03-15" @default.
• W3133996370 creator A5020085889 @default.
• W3133996370 creator A5029033908 @default.
• W3133996370 creator A5035153801 @default.
• W3133996370 date "2021-03-01" @default.
• W3133996370 modified "2023-09-27" @default.
• W3133996370 title "A Little Energy Goes a Long Way: Energy-Efficient, Accurate Conversion from Convolutional Neural Networks to Spiking Neural Networks." @default.
• W3133996370 cites W1604973310 @default.
• W3133996370 cites W1645800954 @default.
• W3133996370 cites W1686810756 @default.
• W3133996370 cites W2020676607 @default.
• W3133996370 cites W2138913040 @default.
• W3133996370 cites W2147800946 @default.
• W3133996370 cites W2156903694 @default.
• W3133996370 cites W2513853720 @default.
• W3133996370 cites W2604319603 @default.
• W3133996370 cites W2775079417 @default.
• W3133996370 cites W2783525259 @default.
• W3133996370 cites W2898323475 @default.
• W3133996370 cites W2949117887 @default.
• W3133996370 cites W2963399222 @default.
• W3133996370 cites W2964338223 @default.
• W3133996370 cites W2989431475 @default.
• W3133996370 cites W3007283957 @default.
• W3133996370 cites W3035644810 @default.
• W3133996370 hasPublicationYear "2021" @default.
• W3133996370 type Work @default.
• W3133996370 sameAs 3133996370 @default.
• W3133996370 citedByCount "0" @default.
• W3133996370 crossrefType "posted-content" @default.
• W3133996370 hasAuthorship W3133996370A5020085889 @default.
• W3133996370 hasAuthorship W3133996370A5029033908 @default.
• W3133996370 hasAuthorship W3133996370A5035153801 @default.
• W3133996370 hasConcept C115961682 @default.
• W3133996370 hasConcept C11731999 @default.
• W3133996370 hasConcept C153180895 @default.
• W3133996370 hasConcept C154945302 @default.
• W3133996370 hasConcept C191178318 @default.
• W3133996370 hasConcept C2776436953 @default.
• W3133996370 hasConcept C41008148 @default.
• W3133996370 hasConcept C50644808 @default.
• W3133996370 hasConcept C81363708 @default.
• W3133996370 hasConceptScore W3133996370C115961682 @default.
• W3133996370 hasConceptScore W3133996370C11731999 @default.
• W3133996370 hasConceptScore W3133996370C153180895 @default.
• W3133996370 hasConceptScore W3133996370C154945302 @default.
• W3133996370 hasConceptScore W3133996370C191178318 @default.
• W3133996370 hasConceptScore W3133996370C2776436953 @default.
• W3133996370 hasConceptScore W3133996370C41008148 @default.
• W3133996370 hasConceptScore W3133996370C50644808 @default.
• W3133996370 hasConceptScore W3133996370C81363708 @default.
• W3133996370 hasLocation W31339963701 @default.
• W3133996370 hasOpenAccess W3133996370 @default.
• W3133996370 hasPrimaryLocation W31339963701 @default.
• W3133996370 hasRelatedWork W2407975742 @default.
• W3133996370 hasRelatedWork W2485907003 @default.
• W3133996370 hasRelatedWork W2552003397 @default.
• W3133996370 hasRelatedWork W2759777444 @default.
• W3133996370 hasRelatedWork W2800920215 @default.
• W3133996370 hasRelatedWork W2884290527 @default.
• W3133996370 hasRelatedWork W2898927016 @default.
• W3133996370 hasRelatedWork W2947957795 @default.
• W3133996370 hasRelatedWork W2981954928 @default.
• W3133996370 hasRelatedWork W3016768016 @default.
• W3133996370 hasRelatedWork W3023721287 @default.
• W3133996370 hasRelatedWork W3128854134 @default.
• W3133996370 hasRelatedWork W3164933491 @default.
• W3133996370 hasRelatedWork W3165298724 @default.
• W3133996370 hasRelatedWork W3175674467 @default.
• W3133996370 hasRelatedWork W3185114446 @default.
• W3133996370 hasRelatedWork W3200038452 @default.
• W3133996370 hasRelatedWork W3205392725 @default.
• W3133996370 hasRelatedWork W3208074141 @default.
• W3133996370 hasRelatedWork W3168557251 @default.
• W3133996370 isParatext "false" @default.
• W3133996370 isRetracted "false" @default.
• W3133996370 magId "3133996370" @default.
• W3133996370 workType "article" @default.