# The Dreaming Variational Autoencoder for Reinforcement Learning Environments

arXiv: https://arxiv.org/abs/1810.01112 · HTML: https://mlsanity.com/paper/1810.01112/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Oct 2018.

## Built on by (most cited first, 4 of 4)

- [Learning to Play Atari in a World of Tokens](https://mlsanity.com/paper/2406.01361/) (2024)
- [Unsupervised Doppler Radar-Based Activity Recognition for e-Healthcare](https://mlsanity.com/paper/2103.10478/) (2021)
- [CostNet: An End-to-End Framework for Goal-Directed Reinforcement Learning](https://mlsanity.com/paper/2210.01805/) (2022)
- [Variational Autoencoders for exteroceptive perception in reinforcement learning-based collision avoidance](https://mlsanity.com/paper/2404.00623/) (2024)

## Abstract (arXiv)

Reinforcement learning has shown great potential in generalizing over raw sensory data using only a single neural network for value optimization. There are several challenges in the current state-of-the-art reinforcement learning algorithms that prevent them from converging towards the global optima. It is likely that the solution to these problems lies in short- and long-term planning, exploration and memory management for reinforcement learning algorithms. Games are often used to benchmark reinforcement learning algorithms as they provide a flexible, reproducible, and easy to control environment. Regardless, few games feature a state-space where results in exploration, memory, and planning are easily perceived. This paper presents The Dreaming Variational Autoencoder (DVAE), a neural network based generative modeling architecture for exploration in environments with sparse feedback. We further present Deep Maze, a novel and flexible maze engine that challenges DVAE in partial and fully-observable state-spaces, long-horizon tasks, and deterministic and stochastic problems. We show initial findings and encourage further work in reinforcement learning driven by generative exploration.