Summit 2018: What I learned from 10 Yrs of "playing" with Neuroevolution - / Al Yaros

Bio-Inspired algorithms such as Genetic Algorithms and Swarm intelligence can be competitive alternative for Training Deep Neural Networks for Reinforcement Learning. My journey in this domain started 10 years ago by building a RL-DNN bot that compete in a 3D racing simulator named Torcs. (demo available on my Youtube channel) - In this talk I'll show a different approach to train a deep neural network that can drive and control a racing car that outperforms a human driver in simulated racing environment - laying the foundations to try and solve any kind of RL problem differently. MP3