Search⌘ K
Home
Courses
A Beginner's Guide to Deep Learning

🍀 Challenge: Train the XOR Multilayer Perceptron

Understand how to train a multilayer perceptron to solve the XOR problem by implementing batch updates through forward propagation, backpropagation, and parameter updates in NumPy. Develop practical skills coding this foundational neural network model and tracking its training loss over epochs.

We'll cover the following...

Problem statement

We have learned that the XOR operator cannot be separated by a line. Therefore, a multilayer perceptron should be used. The following functions implementation is provided below:

  • forward_propagation function
  • backpropagation function
  • update_parameters function
  • calculate_error function

A train function receives the weights, the bias at the two layers respectively, and ...