How Backpropagation Works?

Consider the neural network:
  • two inputs,
  • two hidden neurons,
  • two output neurons, and
  • two biases.
The steps involved in backpropagation are
  1. Forward propagation
  2. Backward propagation
  3. Putting all the values together and calculating the updated weight value
Step 1. Forward Propagation
We will start by propagating forward.


We will repeat this process for the output layer neurons, using the output from the hidden layer neurons as inputs.


Now, let’s see what is the value of the error:





      The boxer is ready to call time (end) on his long career.