TDSNN: From Deep Neural Networks to Deep Spike Neural Networks with Temporal-Coding
Deep Neural Networks
Neural coding
DOI:
10.1609/aaai.v33i01.33011319
Publication Date:
2019-09-13T21:57:19Z
AUTHORS (5)
ABSTRACT
Continuous-valued deep convolutional networks (DNNs) can be converted into accurate rate-coding based spike neural (SNNs). However, the substantial computational and energy costs, which is caused by multiple spikes, limit their use in mobile embedded applications. And recent works have shown that newly emerged temporal-coding SNNs from DNNs reduce load effectively. In this paper, we propose a novel method to convert SNNs, called TDSNN. Combined with characteristic of leaky integrate-andfire (LIF) model, put forward new coding principle Reverse Coding design Ticking Neuron mechanism. According our evaluation, proposed achieves 42% total operations reduction on average large comparing no more than 0.5% accuracy loss. The evaluation shows TDSNN may prove one key enablers make adoption widespread.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (0)
CITATIONS (34)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....