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

The spiking neural network (SNN) computes and communicates informationthrough discrete binary events. It is considered more biologically plausibleand more energy-efficient than artificial neural networks (ANN) in emergingneuromorphic hardware. However, due to the discontinuous and non-differentiablecharacteristics, training SNN is a relatively challenging task. Recent work hasachieved essential progress on an excellent performance by converting ANN toSNN. Due to the difference in information processing, the converted deep SNNusually suffers serious performance loss and large time delay. In this paper,we analyze the reasons for the performance loss and propose a novel bistablespiking neural network (BSNN) that addresses the problem of spikes ofinactivated neurons (SIN) caused by the phase lead and phase lag. Also, whenResNet structure-based ANNs are converted, the information of output neurons isincomplete due to the rapid transmission of the shortcut path. We designsynchronous neurons (SN) to help efficiently improve performance. Experimentalresults show that the proposed method only needs 1/4-1/10 of the time stepscompared to previous work to achieve nearly lossless conversion. We demonstratestate-of-the-art ANN-SNN conversion for VGG16, ResNet20, and ResNet34 onchallenging datasets including CIFAR-10 (95.16% top-1), CIFAR-100 (78.12%top-1), and ImageNet (72.64% top-1).

BSNN: Towards Faster and Better Conversion of Artificial Neural Networks to Spiking Neural Networks with Bistable Neurons