Deep Reinforcement Learning with Double Q-learning

Code

• github.com/labmlai/annotated_deep_learning_paper_implementations
  pytorch★ 66,103
• github.com/facebookresearch/ReAgent
  pytorch★ 3,690
• github.com/opendilab/DI-engine
  pytorch★ 3,606
• github.com/toni-sm/skrl
  jax★ 1,014
• github.com/tensorlayer/RLzoo
  tf★ 644
• github.com/microsoft/med-deadend
  pytorch★ 50
• github.com/OscarHuangWind/Preference-Guided-DQN-Atari
  pytorch★ 12
• github.com/wtingda/DeepRLBreakout
  tf★ 6
• github.com/wmol4/Pytorch_DDQN_Unity_Navigation
  pytorch★ 1
• github.com/jadag/DDQN_mario
  tf★ 0

Abstract

The popular Q-learning algorithm is known to overestimate action values under certain conditions. It was not previously known whether, in practice, such overestimations are common, whether they harm performance, and whether they can generally be prevented. In this paper, we answer all these questions affirmatively. In particular, we first show that the recent DQN algorithm, which combines Q-learning with a deep neural network, suffers from substantial overestimations in some games in the Atari 2600 domain. We then show that the idea behind the Double Q-learning algorithm, which was introduced in a tabular setting, can be generalized to work with large-scale function approximation. We propose a specific adaptation to the DQN algorithm and show that the resulting algorithm not only reduces the observed overestimations, as hypothesized, but that this also leads to much better performance on several games.

Tasks

Benchmark Results

Reproductions