Implicit Quantile Networks for Distributional Reinforcement Learning

Code

• github.com/ku2482/rljax
  jax★ 105
• github.com/robinzixuan/IQN_Agent
  pytorch★ 1
• github.com/sjYoondeltar/myRL_example
  tf★ 0
• github.com/marload/dist-rl-tf2
  tf★ 0
• github.com/SeungeonBaek/discrete-agents-test
  tf★ 0
• github.com/Abdelhamid-bouzid/Distributional-RL
  pytorch★ 0
• github.com/sjYoondeltar/IQN_example
  tf★ 0
• github.com/ACampero/dopamine
  tf★ 0
• github.com/BY571/IQN
  pytorch★ 0
• github.com/ku2482/fqf-iqn-qrdqn.pytorch
  pytorch★ 0

Abstract

In this work, we build on recent advances in distributional reinforcement learning to give a generally applicable, flexible, and state-of-the-art distributional variant of DQN. We achieve this by using quantile regression to approximate the full quantile function for the state-action return distribution. By reparameterizing a distribution over the sample space, this yields an implicitly defined return distribution and gives rise to a large class of risk-sensitive policies. We demonstrate improved performance on the 57 Atari 2600 games in the ALE, and use our algorithm's implicitly defined distributions to study the effects of risk-sensitive policies in Atari games.

Tasks

Benchmark Results

Reproductions