Binary Classification on MNIST (test)
99.9Test AccuracyClassical CNN
Evaluation results
| Method | Links | |
|---|---|---|
| Classical CNNCost Function=Cross-Entropy, Initialization=He Normal, Parameters=∼ 1.2 × 10^5, Training Epochs=20, Convergence State=Global Min2026.03 | 99.9 | |
| Proposed QCNNCost Function=Local (CL), Initialization=TTN Pre-train, Parameters=45, Training Epochs=150, Convergence State=Global Min2026.03 | 98.7 | |
| Baseline QCNNCost Function=Global (CG), Initialization=Random Unif., Parameters=45, Training Epochs=150, Convergence State=Barren Plateau2026.03 | 52.32 |