Binary Classification on MNIST 0 and 1 (test)
100Certification RateAntidote-P
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Antidote-PBias type=MISS, Bias amount (%)=0.05, Perturbation set size=< 10^10, Decision tree depth=22021.10 | 100 | — | — | — | |
| Antidote-PBias type=MISS, Bias amount (%)=0.1, Perturbation set size=< 10^100, Decision tree depth=22021.10 | 100 | — | — | — | |
| Antidote-PBias type=FLIP, Bias amount (%)=0.05, Perturbation set size=< 10^10, Decision tree depth=22021.10 | 100 | — | — | — | |
| Antidote-PBias type=MISS + FAKE, Bias amount (%)=0.05, Perturbation set size=< 10^10, Decision tree depth=22021.10 | 100 | — | — | — | |
| Antidote-PBias type=MISS + FAKE, Bias amount (%)=0.1, Perturbation set size=< 10^100, Decision tree depth=22021.10 | 100 | — | — | — | |
| Antidote-PBias type=MISS + FLIP, Bias amount (%)=0.05, Perturbation set size=< 10^10, Decision tree depth=22021.10 | 100 | — | — | — | |
| Antidote-PBias type=FLIP, Bias amount (%)=0.1, Perturbation set size=< 10^100, Decision tree depth=22021.10 | 95 | — | — | — | |
| Antidote-PBias type=MISS + FLIP, Bias amount (%)=0.1, Perturbation set size=< 10^100, Decision tree depth=22021.10 | 95 | — | — | — | |
| Antidote-PBias type=MISS, Bias amount (%)=0.2, Perturbation set size=< 10^1000, Decision tree depth=22021.10 | 93 | — | — | — | |
| Antidote-PBias type=MISS + FAKE, Bias amount (%)=0.2, Perturbation set size=< 10^1000, Decision tree depth=22021.10 | 93 | — | — | — | |
| Antidote-PBias type=MISS + FLIP, Bias amount (%)=0.2, Perturbation set size=< 10^1000, Decision tree depth=22021.10 | 91 | — | — | — | |
| Antidote-PBias type=MISS, Bias amount (%)=0.4, Perturbation set size=< 10^10000, Decision tree depth=22021.10 | 88 | — | — | — | |
| Antidote-PBias type=FLIP, Bias amount (%)=0.2, Perturbation set size=< 10^1000, Decision tree depth=22021.10 | 88 | — | — | — | |
| Antidote-PBias type=MISS + FAKE, Bias amount (%)=0.4, Perturbation set size=< 10^10000, Decision tree depth=22021.10 | 88 | — | — | — | |
| Antidote-PBias type=MISS + FLIP, Bias amount (%)=0.4, Perturbation set size=< 10^10000, Decision tree depth=22021.10 | 87 | — | — | — | |
| Antidote-PBias type=MISS, Bias amount (%)=0.7, Perturbation set size=> 10^10000, Decision tree depth=22021.10 | 85 | — | — | — | |
| Antidote-PBias type=MISS + FAKE, Bias amount (%)=0.7, Perturbation set size=> 10^10000, Decision tree depth=22021.10 | 85 | — | — | — | |
| Antidote-PBias type=FLIP, Bias amount (%)=0.4, Perturbation set size=< 10^10000, Decision tree depth=22021.10 | 70 | — | — | — | |
| Antidote-PBias type=MISS, Bias amount (%)=1, Perturbation set size=infinite, Decision tree depth=22021.10 | 68 | — | — | — | |
| Antidote-PBias type=MISS + FAKE, Bias amount (%)=1, Perturbation set size=infinite, Decision tree depth=22021.10 | 68 | — | — | — | |
| Antidote-PBias type=MISS + FLIP, Bias amount (%)=0.7, Perturbation set size=> 10^10000, Decision tree depth=22021.10 | 67 | — | — | — | |
| Antidote-PBias type=FLIP, Bias amount (%)=0.7, Perturbation set size=> 10^10000, Decision tree depth=22021.10 | 63 | — | — | — | |
| Antidote-PBias type=MISS + FLIP, Bias amount (%)=1, Perturbation set size=infinite, Decision tree depth=22021.10 | 62 | — | — | — | |
| Antidote-PBias type=FLIP, Bias amount (%)=1, Perturbation set size=infinite, Decision tree depth=22021.10 | 38 | — | — | — | |
| KRRλ=02021.11 | — | 99.95 | 0 | — | |
| KRRλ=0.0012021.11 | — | 99.95 | 0 | — | |
| KRRλ=0.012021.11 | — | 99.95 | 0.0001 | — | |
| KRRλ=0.12021.11 | — | 99.95 | 0.0002 | — | |
| KRRλ=12021.11 | — | 99.95 | 0.001 | — | |
| KRRλ=102021.11 | — | 99.91 | 0.0052 | — | |
| KRRλ=1002021.11 | — | 99.86 | 0.0205 | — | |
| KRRλ=10002021.11 | — | 99.76 | 0.0261 | — | |
| Photonic QCNN (MerLin Optimized)Hyperparameter optimization=Optimized2026.02 | — | 98.8 | — | — | |
| Photonic QCNN (Original)Hyperparameter optimization=Original2026.02 | — | 93.1 | — | — | |
| SVMλ=0.0642021.11 | — | 99.95 | 0.0081 | — |