2020 BRN Discussion, page-27074

When foreign governments start backing research that copies your idea you know you are onto something big. Even they need JAST. My opinion only.

AnSNN Construction Method Based on CNN Conversion and Threshold Setting

YingShang, Yongli Li, Feng You Department of Computer Science, Beijing Universityof Chemical Technology, Beijing, China [email protected],[email protected], [email protected]

Abstract—We present a method to converse the Convolutional Neural Network (CNN) to the Spiking Neural Network (SNN), and we post a threshold adjustment algorithm for SNN. First, the adjustment strategy for CNN is introduced. Then after training, the weight parameters in the model are extracted, which is the corresponding synaptic weight in the layer of the SNN. Finally, a new threshold-setting algorithm based on feedback is proposed to solve the critical problem of the threshold setting of neurons in the SNN. We evaluate our method on the Cifar10 data sets released by Hinton's team. The experimental results show that the image classification accuracy of the SNN is more than 98% of that of CNN, and the theoretical value of power consumption per second is 3.9 mW. Keywords—Convolutional Neural Network; Spiking Neural Network; Threshold Setting Algorithm

ACKNOWLEDGMENTThe National Natural Science Foundation of China under Grant No. 61672085supports the work described in this paper.

References:

[22]Masquelier,T.,&Thorpe,S.J. Unsupervised learning of visual features throughspike timing dependent plasticity. PLoS Computational Biology,3,0247-0257.

https://ksiresearch.org/seke/seke20paper/paper112.pdf