Table 11.

Accuracy and EE benefits for AGD trained SNN with stochmax classifier on VGG5/VGG3x2-v2 topology (refer to Table 9) for different latency T = 25, 50 on CIFAR10 data.

Model	T = 25	T = 50	EE = $\frac{E_{ANN}(1\times )}{E_{SNN}}$	EE = $\frac{E_{SNN(softmax)}}{E_{SNN(stochmax)}}$
	(Accuracy%)		(for T = 25)
VGG5	75.26	75.92	23.83x	1.62x
VGG3x2-v2	72.62	73.17	31.88x	1.97x