From the course: Training Neural Networks in Python

Unlock the full course today

Join today to access over 24,700 courses taught by industry experts.

Solution: Write your own Backpropagation method

Solution: Write your own Backpropagation method - Python Tutorial

From the course: Training Neural Networks in Python

Start my 1-month free trial Buy for my team

Solution: Write your own Backpropagation method

“

- So here's my solution. Step one is the simplest. In line 93, we just run X through the network and assign the result to a new array called outputs. Step two is where we calculate the mean squared error. So first, I save the simple errors in an array called errors, and notice that I'm using NPI vector operations like the subtraction Y minus outputs. Then the mean squared error is the sum of the values in error squared divided by the number of neurons in the last layer. Step three is also done in vector operations. Just following the equation, notice that the result goes to the last element in our D array. Now for step four, pay attention to the loop starting at line 1 0 3. First, I calculate the weighted sum of the forward error terms and then use that sum for the current neurons error term. Notice that the outputs are not recalculated They are fetched from the values cash. All this is assigned to each element in the D…

Contents