Chapter 7: Deep Learning with Neural Networks

Activity 14: Written digit detection

1. This section will discuss how to provide more security for the cryptocurrency traders via the detection of hand-written digits. We will be using assuming that you are a software developer at a new Cryptocurrency trader platform. The latest security measure you are implementing requires the recognition of hand-written digits. Use the MNIST library to train a neural network to recognize digits. You can read more about this dataset on https://www.tensorflow.org/tutorials/.
2. Improve the accuracy of the model as much as possible. And to ensure that it happens correctly, you will need to complete the previous topic.
3. Load the dataset and format the input

   import tensorflow.keras.datasets.mnist as mnist
   (features_train, label_train),
   (features_test, label_test) = mnist.load_data()
   features_train = features_train / 255.0
   features_test = features_test / 255.0
   def flatten(matrix): 
      ...