State representation learning with recurrent capsule networks
This addresses the challenge of state representation learning in reinforcement learning, but appears incremental as it builds on existing capsule network and recurrent methods.
The paper tackles the problem of learning compact and relevant state representations for complex reinforcement learning tasks by proposing a recurrent capsule network that predicts future observations in an agent's trajectory, but no concrete results or numbers are provided.
Unsupervised learning of compact and relevant state representations has been proved very useful at solving complex reinforcement learning tasks. In this paper, we propose a recurrent capsule network that learns such representations by trying to predict the future observations in an agent's trajectory.