English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep spiking neural networks (SNNs) have drawn much attention in recent yearsbecause of their low power consumption, biological rationality and event-drivenproperty. However, state-of-the-art deep SNNs (including Spikformer andSpikingformer) suffer from a critical challenge related to the imprecisegradient backpropagation. This problem arises from the improper design ofdownsampling modules in these networks, and greatly hampering the overall modelperformance. In this paper, we propose ConvBN-MaxPooling-LIF (CML), anSNN-optimized downsampling with precise gradient backpropagation. We prove thatCML can effectively overcome the imprecision of gradient backpropagation from atheoretical perspective. In addition, we evaluate CML on ImageNet, CIFAR10,CIFAR100, CIFAR10-DVS, DVS128-Gesture datasets, and show state-of-the-artperformance on all these datasets with significantly enhanced performancescompared with Spikingformer. For instance, our model achieves 77.64 $\%$ onImageNet, 96.04 $\%$ on CIFAR10, 81.4$\%$ on CIFAR10-DVS, with + 1.79$\%$ onImageNet, +1.16$\%$ on CIFAR100 compared with Spikingformer.

Enhancing the Performance of Transformer-based Spiking Neural Networks by SNN-optimized Downsampling with Precise Gradient Backpropagation