A Comparative Analysis of Different Machine Learning Models to Handwritten Letters and Digits using a Balanced dataset containing 2400 images per class in training, and 400 per class in test set
| Machine Learning Model | Training Accuracy | Test Accuracy |
|---|---|---|
| Naive Bayes | 54.22% | 53.64% |
| Decision Tree | 75.65% | 60.24% |
| Random Forest | 100% | 81.32% |
| Simple Neural Network | 82.20% | 84.22% |
| Convolutional Neural Network | 86.71% | 87.47% |