Deep Neuroevolution: Genetic Algorithms Are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning

arxiv:1712.06567

Original source · Paper JSON

Reinforcement learning · 2017-12-18

Cites (5) · Cited by (4)

Complete cleaned Markdown · Source Markdown including bibliography

Read in parts

1. Part 1
2. Part 2
3. Part 3
4. Part 4

Contents

• Deep Neuroevolution: Genetic Algorithms are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning
• Abstract
• 1 Introduction
• 2 Background
• 3 Methods
• 3.1 Genetic Algorithm
• 3.2 Novelty Search
• 4 Experiments
• 4.1 Atari
• 4.2 Image Hard Maze
• 4.3 Humanoid Locomotion
• 5 Discussion
• 6 Conclusion
• Acknowledgements
• 7 Supplementary Information
• 7.1 Why the GA is faster than ES
• 7.2 Hyperparameters
• 7.3 Additional information about the Deep GA compact encoding method
• 7.4 Additional experimental details for the Image Hard Maze
• 7.5 Humanoid Locomotion
• 7.6 The meaning of “frames”
• Instructions for reporting errors