Steering angle prediction on Udacity (test)
0.17RMSEAsyncSGD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| AsyncSGDAttack Scenario=No attack2026.04 | 0.17 | — | |
| BASGDAttack Scenario=No attack2026.04 | 0.17 | — | |
| Zeno++Attack Scenario=No attack2026.04 | 0.17 | — | |
| SecureAFLAttack Scenario=No attack2026.04 | 0.17 | — | |
| SecureAFLAttack Scenario=Signflip2026.04 | 0.17 | — | |
| SecureAFLAttack Scenario=Min-Max2026.04 | 0.17 | — | |
| SecureAFLAttack Scenario=Adaptive2026.04 | 0.17 | — | |
| KardamAttack Scenario=No attack2026.04 | 0.18 | — | |
| AFLGuardAttack Scenario=No attack2026.04 | 0.18 | — | |
| SecureAFLAttack Scenario=Gaussian2026.04 | 0.18 | — | |
| KardamAttack Scenario=Gaussian2026.04 | 0.19 | — | |
| BASGDAttack Scenario=Signflip2026.04 | 0.19 | — | |
| BASGDAttack Scenario=Gaussian2026.04 | 0.19 | — | |
| AFLGuardAttack Scenario=Min-Max2026.04 | 0.19 | — | |
| BASGDAttack Scenario=Adaptive2026.04 | 0.2 | — | |
| KardamAttack Scenario=Adaptive2026.04 | 0.24 | — | |
| Zeno++Attack Scenario=Gaussian2026.04 | 0.24 | — | |
| AFLGuardAttack Scenario=Gaussian2026.04 | 0.25 | — | |
| AFLGuardAttack Scenario=Adaptive2026.04 | 0.26 | — | |
| KardamAttack Scenario=Signflip2026.04 | 0.28 | — | |
| AsyncSGDAttack Scenario=Signflip2026.04 | 0.29 | — | |
| AsyncSGDAttack Scenario=Adaptive2026.04 | 0.33 | — | |
| AsyncSGDAttack Scenario=Min-Max2026.04 | 0.36 | — | |
| AFLGuardAttack Scenario=Signflip2026.04 | 0.43 | — | |
| KardamAttack Scenario=Min-Max2026.04 | 0.55 | — | |
| AsyncSGDAttack Scenario=Gaussian2026.04 | 1.1 | — | |
| FM-Netbackbone=3D ResNet + LSTM, feature_mimicking=heterogeneous2018.11 | 2.3549 | 1.6236 | |
| 3D ResNet + LSTMbackbone=50-layer ResNet, dimension=3D, temporal_module=LSTM2018.11 | 2.4899 | 1.7147 | |
| 3D CNN + LSTMcomponents=3D CNN, LSTM2018.11 | 2.7167 | 1.8612 | |
| 3D ResNetbackbone=50-layer ResNet, dimension=3D2018.11 | 2.8532 | 1.9167 | |
| 3D CNNcomponents=3D CNN2018.11 | 3.6646 | 2.5598 | |
| CNN + FCNcomponents=CNN, FCN2018.11 | 4.83 | 4.12 | |
| CNN + LSTMcomponents=CNN, LSTM2018.11 | 4.93 | 4.15 | |
| CNN + Attentioncomponents=CNN, Attention2018.11 | 4.93 | 4.15 |