Distributed Species-Based Genetic Algorithm for Reinforcement Learning Problems

EasyChair Preprint 5309

Abstract

Reinforcement Learning (RL) offers a promising solution when dealing with the general problem of optimal decision and control of agents that interact with uncertain environments. A major challenge of existing algorithms is the slow rate of convergence and long training times especially when dealing with high-dimensional state and action spaces. In our work, we leverage evolutionary computing as a competitive alternative to training deep neural networks for RL problems. We present a novel distributed algorithm based on efficient model encoding which enables the intuitive application of genetic operators. Another contribution is the application of crossover operator in two neural networks in the encoded space. Preliminary results demonstrate a considerable reduction of trainable parameters and memory requirements while maintaining comparable performance with DQN and A3C when evaluated on Atari games, resulting in an overall significant speedup.

Keyphrases: Distributed Speciation, Genetic Algorithms, Model encoding, Neuro-evolution strategies, Reinforcement Learning

@booklet{EasyChair:5309,
  author    = {Anirudh Seth and Alexandros Nikou and Marios Daoutis},
  title     = {Distributed Species-Based Genetic Algorithm for Reinforcement Learning Problems},
  howpublished = {EasyChair Preprint 5309},
  year      = {EasyChair, 2021}}