Four different classification models were trained on the famous MNIST dataset, and significant performance metrics were evaluated and compared. Empirical results proved Convoluted Neural Networks outperformed the other classifiers.
The following table summarizes the performance metrics of the four classifiers:
| Classifier | Precision | Recall | F1-Score | Support | Accuracy |
|---|---|---|---|---|---|
| CNN | 0.99 | 0.99 | 0.99 | 10000 | 99.2180% |
| KNN | 0.97 | 0.97 | 0.97 | 14000 | 97.0071% |
| Random Forest | 0.97 | 0.97 | 0.97 | 14000 | 96.7428% |
| SVM | 0.98 | 0.98 | 0.98 | 14000 | 97.6428% |
Please click here