Robust Evolutionary Multi-Objective Network Architecture Search for Reinforcement Learning (EMNAS-RL)

TL;DR

This paper presents EMNAS-RL, a novel evolutionary multi-objective approach using genetic algorithms to optimize neural network architectures for reinforcement learning in autonomous driving, improving rewards and reducing model size.

Contribution

It introduces EMNAS-RL, combining genetic algorithms, parallelization, and transfer learning to automate and enhance neural network architecture search for RL in autonomous driving.

Findings

EMNAS-RL outperforms manually designed models in reward and size.

Parallelization accelerates the search process.

Transfer learning improves learning efficiency and stability.

Abstract

This paper introduces Evolutionary Multi-Objective Network Architecture Search (EMNAS) for the first time to optimize neural network architectures in large-scale Reinforcement Learning (RL) for Autonomous Driving (AD). EMNAS uses genetic algorithms to automate network design, tailored to enhance rewards and reduce model size without compromising performance. Additionally, parallelization techniques are employed to accelerate the search, and teacher-student methodologies are implemented to ensure scalable optimization. This research underscores the potential of transfer learning as a robust framework for optimizing performance across iterative learning processes by effectively leveraging knowledge from earlier generations to enhance learning efficiency and stability in subsequent generations. Experimental results demonstrate that tailored EMNAS outperforms manually designed models, achieving higher rewards with fewer parameters. The findings of these strategies contribute positively to EMNAS for RL in autonomous driving, advancing the field toward better-performing networks suitable for real-world scenarios.