Table 6.
Classification error rate on MNIST and CIFAR-10 dataset.
| Dataset | Architecture | Preprocess | Synap. | ANN | SNN | Error |
|---|---|---|---|---|---|---|
| MNIST | 7-layered ConvNet [ours] | None | 0.33M | 98.5 | 98.5 | 0.0 |
| 7-layered ConvNet (Diehl et al., 2015) | Normalization | 0.33M | 99.14 | 99.12 | 0.02 | |
| CIFAR-10 | AlexNet [ours] | None | 12.98M | 80.23 | 80.12 | 0.11 |
| 8-layered ConvNet (Cao et al., 2015) | Input data preprocessing | 7.4M | 79.12 | 77.43 | 1.69 | |
| 6-layered ConvNet (Rueckauer et al., 2017) | Parameter normalization | 23M | 91.91 | 91.85 | 1.06 | |
| 8-layered Network (Hunsberger et al., 2016) | None | - | 83.72 | 83.54 | 0.18 | |
| VGG-16 [ours] | None | 33M | 88.58 | 88.46 | 0.12 | |
| VGG-16 (Sengupta et al., 2019) | Spiking normalization | - | 91.7 | 91.55 | 0.15 |