Hands-On Neuroevolution with Python

Using NEAT for XOR solver optimization

This was the first chapter in which we started experimenting with the NEAT algorithm. We did this by implementing a solver for one of the classic computer science problems. We started by building a solver for the XOR problem. The XOR problem solver is a computer science experiment in the field of reinforcement learning. The XOR problem cannot be linearly separated and thus requires a solver to find the non-linear execution path. However, we can find the non-linear execution path by introducing hidden layers into the ANN structure.

We discussed how the NEAT algorithm perfectly fits this requirement due to its inherent ability to evolve ANNs from a very simple or a complex topology by gradual complexification. ...