Hands-On Neuroevolution with Python

Deep Neuroevolution

Deep Neural Networks (DNNs) demonstrate outstanding performance improvements in tasks related to pattern recognition and reinforcement learning by utilizing the parallel processing capabilities of modern GPUs. In the context of neuroevolution, it is particularly interesting to explore how conventional deep reinforcement learning (deep RL) methods can be compared to the ones based on Deep Neuroevolution. To answer this question, the research team from UberAI labs created and released the corresponding library in the Python programming language, which uses the TensorFlow framework to handle computations related to neural network training on GPU devices.

The library provides an implementation of the simple genetic algorithm ...