From the course: Training Neural Networks in Python

Unlock the full course today

Join today to access over 24,900 courses taught by industry experts.

The Backpropagation algorithm

The Backpropagation algorithm - Python Tutorial

From the course: Training Neural Networks in Python

Start my 1-month free trial Buy for my team

The Backpropagation algorithm

“

- [Instructor] We are finally ready to see the back propagation training algorithm. This is a general form of the Delta rule. It has several requirements on the neuron model especially on the activation function. Don't worry, the sigmoid makes it easy. The algorithm calculates all weight updates throughout the network. This is done by propagating the error back through the layers. So here are the steps of the backpropagation algorithm to train a multilayer perception with one sample. One, feed a sample to the network. Two, calculate the mean squared error. Three, calculate the error term of each output neuron. Four iteratively calculate the error terms in the hidden layers. Five, apply the delta rule and six, adjust the weights. Now for your next challenge you'll have to write the back propagation algorithm. That's why we're going to have a closed look at each of these steps because it's easy to get confused. For…

Contents