the model model = Sequential() #adding layers to the model model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1))) model.add(MaxPooling2D((2, 2))) model.add(Conv2D(64, (3, 3), activation='relu')) model.add(MaxPooling2D((2, 2))) model.add(Conv2D(64, (3, 3), activation='relu')) #flattening the layer for dense layer connection model.add(Flatten()) #adding a dense layer with softmax activation function model.add(Dense(64, activation='relu')) #hidden layer model.add(Dense(10, activation='softmax')) #output layer § Markdown Compiling the model: The compile step is used to configure the model for training by specifying an optimizer and a loss function and metrics to monitor during training and testing: The Adam optimizer is used here with categorical cross entropy as loss function and accuracy as metric to monitor during training and testing of the model: The learning rate is set to 0.001 which is a good starting point for this problem: § Code #compiling the model with adam optimizer and categorical cross entropy as loss function and accuracy as metric to monitor during training and testing of the model model.compile(optimizer=Adam(), loss='categorical_crossentropy', metrics=['accuracy']) § Markdown Training the model: The fit() method on our model takes numpy arrays of input data (X_train) and target data (y_train) as arguments and trains our network in batches using stochastic gradient descent: We also specify number of epochs which is set to 10 for this problem which means that we are going through our dataset 10 times in order to train our network: We also specify validation data which will be used for evaluating our network after each epoch: We also specify verbose argument which will be used for printing out progress of our training process: Finally we can also specify callbacks argument which can be used for performing certain actions after each epoch such as saving weights of our network or early stopping in case of overfitting etc.: After fitting our model we can evaluate it on our test data in order to get an idea about how well it is performing on unseen data: We can also use predict() method on our trained model in order to make predictions on new data points: Finally we can save weights of our trained network using save_weights() method so that we don't have to train it again if we want to use it later on some other task or same task but with different data points etc.: Training the Model : The fit() method on our model takes numpy arrays of input data (X_train) and target data (y_train) as arguments and trains our network in batches using stochastic gradient descent : We also specify number of epochs which is set to 10 for this problem which means that we are going through our dataset 10 times in order to train our network : We also specify validation data which will be used for evaluating our network after each epoch : We also specify verbose argument which will be used for printing out progress of our training process : Finally we can also specify callbacks argument which can be used for performing certain actions after each epoch such as saving weights of our network or early stopping in case of overfitting etc . : After fitting our model we can evaluate it on our test data in order to get an idea about how well it is performing on unseen data : We can also use predict() method on