Research Article
Auxiliary Pneumonia Classification Algorithm Based on Pruning Compression
Table 2
Comparison of CP method with other network models at different pruning rates.
| Model | Test acc (%) | Params (M) | MAdd (G) | FLOPs (G) |
| CPResNet50 (50% pruned) | 98.189 | 10.337 | 3.68 | 1.85 | CPResNet50 (70% pruned) | 97.707 | 6.682 | 2.4 | 1.21 | CPResNet50 (90% pruned) | 97.888 | 3.888 | 1.38 | 0.696 | VGG16 [36] | 96.619 | 14.773 | 30.77 | 15.41 | DenseNet121 [37] | 97.948 | 6.956 | 5.74 | 2.88 | GoogLeNet [38] | 98.371 | 5.60 | 3.02 | 3.02 | Inception_v3 [39] | 93.784 | 21.79 | 5.69 | 2.85 |
|
|