Reinforcement learning is a machine learning paradigm in which an agent, situated in an environment, must resolve a task through repeated experimentation. In the last decade, the combination of reinforcement learning with deep learning has given rise to deep reinforcement learning (DRL), a family of algorithms that use neural networks to select the best agent’s actions to achieve its goal. Deep reinforcement learning has proved highly effective in solving problems in different fields, such as robotics, strategy games and industrial control. Among other milestones, deep reinforcement learning algorithms have been able to defeat professional Go players, predict 3D protein models and operate robots in unknown environments.

The first Workshop on Deep Reinforcement Learning (DeRL 2021) aims to bring together researchers interested in learning about and sharing the latest advances in this area, both theoretical and applied.




Topics of interest for DeRL include (but are not limited to) the following:

  • DRL algorithms
  • Advanced neural architectures in DRL
  • Frameworks and simulation environments for DRL
  • Meta-learning in DRL
  • State representation learning
  • Sparse rewards and few-shot learning
  • Curriculum and continual learning
  • Imprecision and uncertainty in DRL
  • Explainable DRL
  • Multi-agent DRL
  • Applications of DRL: control, games, robotics, etc.

Waiting for your contributions!

Download the CfP