The following steps represent step-by-step description of how information goes forward in a neural network. This process is referred to as forward propagation:
- The input values arrive at the input layer and are processed in the neurons.
- The outputs are then forwarded to the hidden layers wherein the randomly initialized weights are multiplied by the values and the bias is added.
- These values are then passed through the activation function.
- Finally, the values reach the output layer and the neurons perform the processing and emit an output value, y'.
- This y' is the predicted value for the input that came in.
Everything that we have discussed hitherto falls under the category of forward propagation.
As we saw, a value y' was predicted by the network. No learning has happened yet.
Now we need to judge the performance of our network, in terms of how far away or close it was to predicting the correct value. We do this by measuring something...
