Neural Networks – Matrices and Initializers

Matrices

This is a function we need to specify to a the keras.compile function to measure the output accuracy of an NN. https://keras.io/metrics/

This has nothing to do with how the NN works but rather to measure the final output accuracy.

Read more here https://stackoverflow.com/questions/34518656/how-to-interpret-loss-and-accuracy-for-a-machine-learning-model

To complete our NN, we finally use this accuracy as follows

model.compile(Adam(lr=.0001),loss="sparse_categorical_crossentropy",metrics=['accuracy'])
# this completes our first neural network

Initializers

These are helper function which are optional to use in case we need to initialize our NN weights with some specific values instead of total random values. https://keras.io/initializers/

e.g

model.add(Dense(64,
                kernel_initializer='random_uniform',
                bias_initializer='zeros'))
excellence-social-linkdin
excellence-social-facebook
excellence-social-instagram
excellence-social-skype