MNIST Digit Classification
The MNIST dataset is a classic image classification problem, consisting of 28x28 images of handwritten digits 0-9. The intent of this project is to build a neural network using Keras that predicts the digit with an accuracy of 95% or better. My personal grade scale is as follows:
| Validation Accuracy | Grade |
|---|---|
| A+ | >99% |
| A | 95-99% |
| B | 80-95% |
| C | 70-79% |
| D | 60-69%% |
| F | <60% |