Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204249096> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W3204249096 endingPage "4680" @default.
- W3204249096 startingPage "4671" @default.
- W3204249096 abstract "Spiking Neural Networks (SNNs) offer a promising alternative to traditional deep learning frameworks, since they provide higher computational efficiency due to event-driven information processing. SNNs distribute the analog values of pixel intensities into binary spikes over time. However, the most widely used input coding schemes, such as Poisson based rate-coding, do not leverage the additional temporal learning capability of SNNs effectively. Moreover, these SNNs suffer from high inference latency which is a major bottleneck to their deployment. To overcome this, we propose a scalable time-based encoding scheme that utilizes the Discrete Cosine Transform (DCT) to reduce the number of timesteps required for inference. DCT decomposes an image into a weighted sum of sinusoidal basis images. At each time step, a single frequency base, taken in order and modulated by its corresponding DCT coefficient, is input to an accumulator that generates spikes upon crossing a threshold. We use the proposed scheme to learn DCT-SNN, a low-latency deep SNN with leaky-integrate-and-fire neurons, trained using surrogate gradient descent based backpropagation. We achieve top-1 accuracy of 89.94%, 68.3% and 52.43% on CIFAR-10, CIFAR-100 and TinyImageNet, respectively using VGG architectures. Notably, DCT-SNN performs inference with 2-14X reduced latency compared to other state-of-the-art SNNs, while achieving comparable accuracy to their standard deep learning counterparts. The dimension of the transform allows us to control the number of timesteps required for inference. Additionally, we can trade-off accuracy with latency in a principled manner by dropping the highest frequency components during inference." @default.
- W3204249096 created "2021-10-11" @default.
- W3204249096 creator A5031161187 @default.
- W3204249096 creator A5066407038 @default.
- W3204249096 creator A5081219427 @default.
- W3204249096 date "2021-05-04" @default.
- W3204249096 modified "2023-09-26" @default.
- W3204249096 title "DCT-SNN: Using DCT to Distribute Spatial Information over Time for Learning Low-Latency Spiking Neural Networks" @default.
- W3204249096 cites W1604973310 @default.
- W3204249096 cites W1645800954 @default.
- W3204249096 cites W1999085092 @default.
- W3204249096 cites W2020676607 @default.
- W3204249096 cites W2031614119 @default.
- W3204249096 cites W2040903332 @default.
- W3204249096 cites W2160815625 @default.
- W3204249096 cites W2163605009 @default.
- W3204249096 cites W2541249978 @default.
- W3204249096 cites W2584959338 @default.
- W3204249096 cites W2612445135 @default.
- W3204249096 cites W2775079417 @default.
- W3204249096 cites W2806066966 @default.
- W3204249096 cites W2904620099 @default.
- W3204249096 cites W2962908174 @default.
- W3204249096 cites W2963760575 @default.
- W3204249096 cites W2964296416 @default.
- W3204249096 cites W2969335882 @default.
- W3204249096 cites W2972545288 @default.
- W3204249096 cites W2987748894 @default.
- W3204249096 cites W3016620475 @default.
- W3204249096 cites W3023721287 @default.
- W3204249096 cites W3102087395 @default.
- W3204249096 cites W3121127024 @default.
- W3204249096 cites W3159754263 @default.
- W3204249096 cites W2006370340 @default.
- W3204249096 hasPublicationYear "2021" @default.
- W3204249096 type Work @default.
- W3204249096 sameAs 3204249096 @default.
- W3204249096 citedByCount "0" @default.
- W3204249096 crossrefType "proceedings-article" @default.
- W3204249096 hasAuthorship W3204249096A5031161187 @default.
- W3204249096 hasAuthorship W3204249096A5066407038 @default.
- W3204249096 hasAuthorship W3204249096A5081219427 @default.
- W3204249096 hasConcept C108583219 @default.
- W3204249096 hasConcept C115961682 @default.
- W3204249096 hasConcept C11731999 @default.
- W3204249096 hasConcept C153180895 @default.
- W3204249096 hasConcept C154945302 @default.
- W3204249096 hasConcept C2221639 @default.
- W3204249096 hasConcept C2776214188 @default.
- W3204249096 hasConcept C41008148 @default.
- W3204249096 hasConcept C50644808 @default.
- W3204249096 hasConceptScore W3204249096C108583219 @default.
- W3204249096 hasConceptScore W3204249096C115961682 @default.
- W3204249096 hasConceptScore W3204249096C11731999 @default.
- W3204249096 hasConceptScore W3204249096C153180895 @default.
- W3204249096 hasConceptScore W3204249096C154945302 @default.
- W3204249096 hasConceptScore W3204249096C2221639 @default.
- W3204249096 hasConceptScore W3204249096C2776214188 @default.
- W3204249096 hasConceptScore W3204249096C41008148 @default.
- W3204249096 hasConceptScore W3204249096C50644808 @default.
- W3204249096 hasLocation W32042490961 @default.
- W3204249096 hasOpenAccess W3204249096 @default.
- W3204249096 hasPrimaryLocation W32042490961 @default.
- W3204249096 hasRelatedWork W1991618012 @default.
- W3204249096 hasRelatedWork W2283854151 @default.
- W3204249096 hasRelatedWork W2295423501 @default.
- W3204249096 hasRelatedWork W2305996987 @default.
- W3204249096 hasRelatedWork W2806066966 @default.
- W3204249096 hasRelatedWork W2963966976 @default.
- W3204249096 hasRelatedWork W2990498047 @default.
- W3204249096 hasRelatedWork W2996615169 @default.
- W3204249096 hasRelatedWork W3004073022 @default.
- W3204249096 hasRelatedWork W3004676899 @default.
- W3204249096 hasRelatedWork W3032298043 @default.
- W3204249096 hasRelatedWork W3034194585 @default.
- W3204249096 hasRelatedWork W3044912338 @default.
- W3204249096 hasRelatedWork W3045777102 @default.
- W3204249096 hasRelatedWork W3047826181 @default.
- W3204249096 hasRelatedWork W3104409553 @default.
- W3204249096 hasRelatedWork W3110176854 @default.
- W3204249096 hasRelatedWork W3124546450 @default.
- W3204249096 hasRelatedWork W3201524232 @default.
- W3204249096 hasRelatedWork W3213073654 @default.
- W3204249096 isParatext "false" @default.
- W3204249096 isRetracted "false" @default.
- W3204249096 magId "3204249096" @default.
- W3204249096 workType "article" @default.