Skip to main content
. 2019 Dec 14;19(24):5541. doi: 10.3390/s19245541

Table 6.

The improvements of the proposed best model for the CIFAR-10 data set over the state-of-the-art networks, grouped by network sizes.

Model Acc Improvement FLOPs Saving
EffNet V1 large 4.88% 87.59%
MobileNet large 11.72% 14.66%
ShuffleNet large 12.0% 10.81%
EffNet V1 9.7% 13.16%
EffNet V2 8.23% 45.3%
MobileNetV2 13.43% 39.63%
MobileNet 9.71% 27.59%
ShuffleNet 9.89% 10.64%