1

The goal here is from a built keras model to find the coefficient between each neurons from the information gathered get_weights()

I'm building the system as a class so it may be easily implemented within different projects so as a test model I've used the 'mnist' example provided on the tensor flow main learn page.

Please correct me if I'm wrong, but I've gathered that from the model below:

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28)),
  tf.keras.layers.Dense(512, activation=tf.nn.relu),
  tf.keras.layers.Dropout(0.2),
  tf.keras.layers.Dense(10, activation=tf.nn.softmax)
])

The set of arrays/matrixes retrieved from 'model.get_weights()' correspond solely to the model's dense layer. Gathered from overflow post linked below, each dense layer contributes a matrix of weights coming into the layer and a layer of biases for each neuron. Overflow post: Keras: Interpreting the output of get_weights()

When it comes to calculating the coefficient between two neurons I haven't found much details in regards a technical definition of what is needed to find said coefficient.

Between calculating the coefficient between the neurons, I'm looking for some help or direction to where to look for a detailed equations or explanations to solving for the numerical answer.

Thank you. -M

Malmsten_
  • 11
  • 1
  • Maybe you want to elaborate on what you mean by the "coefficient between the neurons". Perhaps if you state your goal it may be easier for people to point you in the right direction. – Pedro Marques Jul 20 '19 at 10:23

0 Answers0