Research Article
Robustness and Explainability of Image Classification Based on QCNN
Table 2
QCNN accuracy against FGSM adversarial attacks.
| Parameter | 32/255 | 24/255 | 16/255 | 12/255 |
| FGSMs | 0.9062 | 0.9219 | 0.9487 | 0.9844 | Bitplane1 | 0.7394 | 0.7439 | 0.7426 | 0.7541 | Bitplane2 | 0.7278 | 0.7297 | 0.7384 | 0.7641 | Bitplane3 | 0.7386 | 0.7344 | 0.7528 | 0.7921 | Bitplane4 | 0.7473 | 0.7469 | 0.7531 | 0.7853 | Bitplane5 | 0.9062 | 0.8281 | 0.8594 | 0.8750 | Bitplane6 | 0.8906 | 0.9688 | 0.8906 | 0.8281 | Bitplane7 | 0.9844 | 0.9665 | 0.8750 | 0.9688 | Bitplane8 | 0.9375 | 0.9531 | 0.9219 | 0.9844 |
|
|