Three-class classification on Stroke Dataset (subject-wise cross-validation)
83Macro F1 ScoreES-VAE + k-NN
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ES-VAE + k-NNInput Representation=Tangent Vector2026.05 | 83 | 86 | 80 | |
| Tangent PCA + k-NNInput Representation=Tangent Vector2026.05 | 79 | 83 | 76 | |
| TCNInput Representation=Tangent Vector2026.05 | 75 | 78 | 74 | |
| LSTMInput Representation=Tangent Vector2026.05 | 70 | 81 | 67 | |
| VAE + k-NNInput Representation=Joint Angle2026.05 | 69 | 78 | 65 | |
| TCNInput Representation=Raw Skeleton2026.05 | 64 | 70 | 62 | |
| TransformerInput Representation=Raw Skeleton2026.05 | 63 | 63 | 63 | |
| TransformerInput Representation=Tangent Vector2026.05 | 63 | 75 | 60 | |
| VAE + k-NNInput Representation=Raw Skeleton2026.05 | 61 | 68 | 58 | |
| PCA + k-NNInput Representation=Joint Angle2026.05 | 61 | 78 | 57 | |
| LSTMInput Representation=Raw Skeleton2026.05 | 56 | 57 | 56 | |
| Hyper-GCNInput Representation=Tangent Vector2026.05 | 56 | 56 | 57 | |
| PCA + k-NNInput Representation=Raw Skeleton2026.05 | 55 | 55 | 55 | |
| Sparse ST-GCNInput Representation=Tangent Vector2026.05 | 53 | 73 | 49 | |
| Sparse ST-GCNInput Representation=Raw Skeleton2026.05 | 48 | 75 | 49 | |
| ST-GCNInput Representation=Tangent Vector2026.05 | 48 | 50 | 48 | |
| Hyper-GCNInput Representation=Raw Skeleton2026.05 | 47 | 48 | 47 | |
| ST-GCNInput Representation=Raw Skeleton2026.05 | 39 | 48 | 40 |