Cleaned up version of NEAT code used in the paper: "Data-Efficient Neuroevolution with Kernel-Based Surrogate Models" (https://arxiv.org/abs/1804.05364)
The surrogate-assistance/Gaussian process code has been stripped out for those interested only in a MATLAB version of NEAT. It includes experiments for XOR, double pole balancing, cart-pole swing up, and half-cheetah (with MuJoCo and OpenAI gym).
See usage examples in 'demo.m' for, um, examples of usage.