Build and Fit the Model A

Create a keras neural network as follows:

• Add keras.layers.Flatten to flatten the input image to the model.

• Add 5 dense layers with n_hidden number of neurons and selu activation function.

• Add a final dense layer with 8 neurons and softmax activation function(for classifying 8 classes of data).

  model_A = keras.models.Sequential()
  model_A.add(keras.layers.Flatten(input_shape=[28, 28]))
  for n_hidden in (300, 100, 50, 50, 50):
      model_A.add(keras.layers.Dense(n_hidden, activation="selu"))
  model_A.add(keras.layers.Dense(8, activation="softmax"))
  

Define the compiling part of the model:

• Set "sparse_categorical_crossentropy" as loss.

• Set keras.optimizers.SGD(lr=1e-3) as optimizer.

  model_A.compile(loss= << your code comes here >>,
      optimizer= << your code comes here >>,
      metrics=["accuracy"])
  

Now train model_A on X_train_A and y_train_A for 5 epochs , and validation_data=(X_valid_A, y_valid_A) using model_A.fit.

history = << your code comes here >>(X_train_A, y_train_A, epochs=5,
            validation_data=(X_valid_A, y_valid_A))

Save the model_A we created.

model_A.save("my_model_A.h5")