January 2018
Beginner to intermediate
284 pages
8h 35m
English
With the errors, backpropagation works backwards to update the network weights on the direction of gradients of the error. First, we need to compute the deltas of the weights and biases. Note that
is used to update
and
, and
is used to update and :
This is written in TensorFlow code as follows:
d_z_2 = tf.multiply(error, sigmoidprime(z_2)) ...
Read now
Unlock full access