Classification on Radar Signal Processing Dataset (5-fold CV)
98.4AccuracyCorNet-LECM
Evaluation Results
| Method | Links | |
|---|---|---|
| CorNet-LECMManifold=Correlation, Training time per epoch=1.122026.05 | 98.4 | |
| CorNet-ECMManifold=Correlation, Training time per epoch=1.012026.05 | 97.71 | |
| CorNet-OLMManifold=Correlation, Training time per epoch=1.352026.05 | 97.57 | |
| C-SPMEvaluation Architecture=Linear Probe2026.05 | 96.87 | |
| LCEvaluation Architecture=Linear Probe2026.05 | 96.77 | |
| θ-PCMEvaluation Architecture=Linear Probe, theta=0.52026.05 | 96.6 | |
| S-SPMEvaluation Architecture=SPD MLR2026.05 | 96.57 | |
| CorNet-PHCMManifold=Correlation, Training time per epoch=2.372026.05 | 96.56 | |
| ALEM (Learnable)Evaluation Architecture=Linear Probe2026.05 | 96.5 | |
| S-SPMEvaluation Architecture=Linear Probe2026.05 | 96.33 | |
| LEEvaluation Architecture=Linear Probe2026.05 | 96.3 | |
| GyroAI*Manifold=SPD, Training time per epoch=0.992026.05 | 96.29 | |
| AIRMEvaluation Architecture=Linear Probe2026.05 | 96.27 | |
| AIRMEvaluation Architecture=SPD MLR2026.05 | 96.27 | |
| GyroLE*Manifold=SPD, Training time per epoch=0.792026.05 | 96.24 | |
| CorNet-LSMManifold=Correlation, Training time per epoch=1.502026.05 | 96.24 | |
| LCEvaluation Architecture=SPD MLR2026.05 | 96.17 | |
| C-SPMEvaluation Architecture=SPD MLR2026.05 | 96.17 | |
| LEEvaluation Architecture=SPD MLR2026.05 | 96 | |
| θ-PCMEvaluation Architecture=SPD MLR, theta=0.52026.05 | 95.93 | |
| SPDResNet-LEMManifold=SPD, Training time per epoch=0.772026.05 | 95.89 | |
| ALEM (Learnable)Evaluation Architecture=SPD MLR2026.05 | 95.8 | |
| SPDResNet-AIMManifold=SPD, Training time per epoch=0.962026.05 | 95.71 | |
| SPDNetLieBN-AIMManifold=SPD, Training time per epoch=1.212026.05 | 95.47 | |
| GyroSPD++*Manifold=SPD, Training time per epoch=5.092026.05 | 95.2 | |
| SPDNetBNManifold=SPD, Training time per epoch=1.252026.05 | 94.85 | |
| SPDNetLieBN-LCMManifold=SPD, Training time per epoch=1.102026.05 | 94.8 | |
| SPDNetMLRManifold=SPD, Training time per epoch=0.662026.05 | 94.59 | |
| GyroLC*Manifold=SPD, Training time per epoch=0.662026.05 | 93.6 | |
| SPDNetManifold=SPD, Training time per epoch=0.662026.05 | 93.25 | |
| S-SPMEvaluation Architecture=RResNet2026.05 | 93.23 | |
| LCEvaluation Architecture=RResNet2026.05 | 93.03 | |
| LEEvaluation Architecture=RResNet2026.05 | 92.67 | |
| C-SPMEvaluation Architecture=RResNet2026.05 | 92.4 | |
| ALEM (Learnable)Evaluation Architecture=RResNet2026.05 | 92.1 | |
| θ-PCMEvaluation Architecture=RResNet, theta=-1.02026.05 | 91.83 | |
| AIRMEvaluation Architecture=RResNet2026.05 | 90.87 | |
| GyroGr*Manifold=Grassmann, Training time per epoch=1.382026.05 | 90.64 | |
| GrNetManifold=Grassmann, Training time per epoch=1.392026.05 | 90.48 | |
| GyroGr-Scaling*Manifold=Grassmann, Training time per epoch=1.632026.05 | 88.88 |