After the struggles I faced coding my own NEAT algorithm to beat the game Breakout, I've decided to take a step back and research what makes a successful NEAT algorithm by using OpenAI Gym Retro and a pure Python implementation of NEAT called NEAT-Python. I've used a Python script to connect the two. To understand NEAT, you have to understand where it comes from - its roots. It falls under a larger category called TWEANNs. This stands for Topology and Weight Evolving Artificial Neural Networks. These are algorithms not only evolve the strength of the connection weights for a fixed network topology, but actually evolves both the topology of the network and its weights. The folks behind the NEAT algorithm identified three major challenges for TWEANNs: meaningful crossover, speciation, and structure complexity. They designed NEAT to specifically address each one by tracking genes through historical markings, protecting structural innovation through speciation, and incrementally growing from minimal structure, respectively. This prevents the algorithm from blindly crossing over the genomes of two neural networks and creating horribly mutated networks (crossover). It protects new structures as they are typically at a disadvantage, allowing them to optimize before we eliminate them. This is done by splitting up the population into several species based on the similarity of topology and connections and only competing within their species (speciation). And it prevents the algorithm from creating complex networks that may have to later reduce the number of nodes and connections after the fact. They did this by starting all networks with no hidden nodes - it only has the input nodes, output nodes, and a series of connection genes between them. Then it can evolve in complexity if and only if it is found to be useful and necessary (minimal structure). If you found this video helpful or entertaining, please leave a like & subscribe! 🐱‍🚀 GitHub: https://github.com/forrestknight 🐦 Twitter: https://www.twitter.com/forrestpknight 📸 Instagram: https://www.instagram.com/forrestpknight 👨‍💻 My Coding Gear: https://bit.ly/my-coding-gear 📓 Learning Resources: My Favorite Machine Learning Course: https://imp.i384100.net/YgYEBJ Open Source Computer Science Degree: https://bit.ly/open-source-forrest Udacity to Learn Any Coding Skill: http://bit.ly/udacity-forrest 🔧Coding Tools: The Best Linux Server Hosting: https://bit.ly/linode-forrest