Table 3.
Classification performance (%) comparison using different approaches for the DRIVE dataset.
| Approach/Metrics | AUC | ACC |
|---|---|---|
| Generative Models | ||
| Gaussian Mixture | 0.70 | 84.01 |
| Dirichlet Mixture | 0.72 | 84.79 |
| Scaled Dirichlet Mixture | 0.75 | 84.99 |
| Shifted Scaled Dirichlet Mixture | 0.77 | 85.36 |
| Hybrid Models | ||
| Gaussian Mixture + Fisher Kernel | 0.81 | 87.84 |
| Gaussian Mixture + Bhattacharyya Kernel | 0.81 | 89.02 |
| Gaussian Mixture + Kullback–Leibler Kernel | 0.81 | 87.11 |
| Dirichlet Mixture + Fisher Kernel | 0.84 | 88.54 |
| Dirichlet Mixture + Bhattacharyya Kernel | 0.86 | 90.67 |
| Dirichlet Mixture + Kullback–Leibler Kernel | 0.84 | 88.01 |
| Scaled Dirichlet Mixture + Fisher Kernel | 0.87 | 90.87 |
| Scaled Dirichlet Mixture + Bhattacharyya Kernel | 0.90 | 91.33 |
| Scaled Dirichlet Mixture + Kullback–Leibler Kernel | 0.85 | 88.14 |
| Shifted Scaled Dirichlet Mixture + Fisher Kernel | 0.88 | 91.13 |
| Shifted Scaled Dirichlet Mixture + Bhattacharyya Kernel | 0.91 | 91.65 |
| Shifted Scaled Dirichlet Mixture + Kullback–Leibler Kernel | 0.91 | 88.98 |
| Other Methods | ||
| Fleming et al. [61] | 89.80 | |
| Garcia et al. [62] | 73.55 | |
| Li and Chutatape [63] | 85.50 | |
| Wang et al. [64] | 85.00 |