Table 1 Comparison of the implementation of a basic neural computation with n input neurons and 1 output neuron.
From: A co-design framework of neural networks and quantum circuits towards quantum advantage
Layers | FC(C)58 | FC(Q)31 | P-LYR | U-LYR | |
|---|---|---|---|---|---|
Space | O(n) (or O(1)) | \(O({\mathrm{log}}\,n)\) | O(n) | \(O({\mathrm{log}}\,n)\) | |
Complexity | Time | O(1) (or O(n)) | O(n) | \(O(n\cdot \mathrm{log}\,n)\) | \(O({\mathrm{log}\,}^{2}n)\) |
Cost | O(n) | \(O(n\cdot {\mathrm{log}}\,n)\) | \(O({n}^{2}\cdot {\mathrm{log}}\,n)\) | \(O({\mathrm{log}\,}^{3}n)\) | |
Data type | Input data | F32 | Bin | R.V. | F32 |
Weights | Bin (F32) | Bin | Bin (R.V.) | Bin | |
Conn. w/o measurement | ✓ | – | ✓ | × | |
Summary | Flexibility | – | × | ✓ | × |
Qu. Adv. | – | × | × | ✓ | |
