EasyDeepLearn

Dyna — early model-based RL framework.

medium

Answer

  • Sutton 1990: interleave (1) real interaction with env, (2) model updates from real data, (3) planning: multiple simulated Q-learning updates from the learned model.
  • Rebuilds Q from cheap simulated experience.
  • Cornerstone of model-based RL.
  • Modern deep versions: Dyna-Q with NN model, MBPO (Janner et al. 2019).
  • Advantage: model-based sample efficiency + model-free performance.
Check yourself — multiple choice
  • Random
  • Interleave real interaction + model update + planning via simulated Q-updates from learned model; cornerstone of MBRL; modern: MBPO / Dyna-Q with NN
  • Same as DQN
  • Not real

Dyna: real + model + planning simulated updates; MBRL foundation.

#model-based#planning

Practise Reinforcement Learning

214 interview questions in this topic.

Related questions