Self-Organisation of Neural Topologies by Evolutionary Reinforcement Learning

  • Nils T. Siebel
  • Jochen Krause
  • Gerald Sommer
Schlagworte: Neural Networks, Evolutionary Algorithms, Reinforcement Learning, DDC: 004 (Data processing, computer science, computer systems)


In this article we present EANT, "Evolutionary Acquisition of Neural Topologies", a method that creates neural networks (NNs) by evolutionary reinforcement learning. The structure of NNs is developed using mutation operators, starting from a minimal structure. Their parameters are optimised using CMA-ES. EANT can create NNs that are very specialised; they achieve a very good performance while being relatively small. This can be seen in experiments where our method competes with a different one, called NEAT, "NeuroEvolution of Augmenting Topologies", to create networks that control a robot in a visual serving scenario.