Evolutionary Planning in Latent Space

TL;DR

This paper introduces a method for reinforcement learning that combines learned latent space representations with evolutionary planning, enabling effective decision-making without requiring an explicit world model.

Contribution

The paper presents a novel framework called EPLS that learns a stochastic world model in latent space and uses evolutionary algorithms for planning, improving over standard model-free methods.

Findings

EPLS outperforms standard model-free reinforcement learning agents.

The learned world model enables effective planning in complex environments.

Iterative refinement improves the accuracy of the world model and planning performance.

Abstract

Planning is a powerful approach to reinforcement learning with several desirable properties. However, it requires a model of the world, which is not readily available in many real-life problems. In this paper, we propose to learn a world model that enables Evolutionary Planning in Latent Space (EPLS). We use a Variational Auto Encoder (VAE) to learn a compressed latent representation of individual observations and extend a Mixture Density Recurrent Neural Network (MDRNN) to learn a stochastic, multi-modal forward model of the world that can be used for planning. We use the Random Mutation Hill Climbing (RMHC) to find a sequence of actions that maximize expected reward in this learned model of the world. We demonstrate how to build a model of the world by bootstrapping it with rollouts from a random policy and iteratively refining it with rollouts from an increasingly accurate planning policy using the learned world model. After a few iterations of this refinement, our planning agents are better than standard model-free reinforcement learning approaches demonstrating the viability of our approach.