NN architecture:
- sequential
NN layers:
- input layer with 28*28 nodes (x_train_reshaped.shape)
- two hidden layers:
- 1st layer has 128 nodes
- 2nd layer has 64 nodes
- output layer with 10 nodes (y_train_cat.shape)
- used dropout on hidden layers to reduce overfitting
Epochs:
- 10 epochs
Accuracy:
- 95.15%