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

In recent years, studies on image generation models of spiking neuralnetworks (SNNs) have gained the attention of many researchers. Variationalautoencoders (VAEs), as one of the most popular image generation models, haveattracted a lot of work exploring their SNN implementation. Due to theconstrained binary representation in SNNs, existing SNN VAE methods implicitlyconstruct the latent space by an elaborated autoregressive network and use thenetwork outputs as the sampling variables. However, this unspecified implicitrepresentation of the latent space will increase the difficulty of generatinghigh-quality images and introduces additional network parameters. In thispaper, we propose an efficient spiking variational autoencoder (ESVAE) thatconstructs an interpretable latent space distribution and design areparameterizable spiking sampling method. Specifically, we construct the priorand posterior of the latent space as a Poisson distribution using the firingrate of the spiking neurons. Subsequently, we propose a reparameterizablePoisson spiking sampling method, which is free from the additional network.Comprehensive experiments have been conducted, and the experimental resultsshow that the proposed ESVAE outperforms previous SNN VAE methods inreconstructed & generated images quality. In addition, experiments demonstratethat ESVAE's encoder is able to retain the original image information moreefficiently, and the decoder is more robust. The source code is available athttps://github.com/QgZhan/ESVAE.

ESVAE: An Efficient Spiking Variational Autoencoder with Reparameterizable Poisson Spiking Sampling