Solving the XOR Problem with a Feedforward Neural Network
In the course of a corporate project, there always comes the point when a problem that seems impossible to solve hits you. At that point, you try everything you've learned, but it doesn't work for what's asked of you. Your team or customer begins to look elsewhere. It's time to react.
In this chapter, an impossible-to-solve business case regarding material optimization will be resolved successfully with a hand-made version of a feedforward neural network (FNN) with backpropagation.
Feedforward networks are one of the key building blocks of deep learning. The battle around the XOR function perfectly illustrates how deep learning regained popularity in corporate environments. XOR is an exclusive OR function that we will explore later in this chapter. The XOR FNN illustrates one of the critical functions of neural networks: classification. Once information becomes classified into subsets...
