Spiking Deep Learning with SpikingJelly

Presentation type: Invited Talk

Abstract

With deep learning methods being applied, SNNs get huge promotions on performance and become the promising potential competitor of ANNs. However, the software frameworks for deep learning with SNNs are still underdeveloped, which increases the workload of researchers and hinders models/codes sharing. To promote the community, we propose the SpikingJelly, an open-source deep learning framework for SNNs. In this talk, we will introduce the core design conception and modules of SpikingJelly, showing how SpikingJelly achieves extremely high efficiency with flexible and handy APIs. Then, we will introduce research from our groups and show how to use SpikingJelly to promote the spiking deep learning.