EasyDeepLearn

How does posterior sampling drive exploration in deep RL?

hard

Answer

  • Maintain posterior over Q or model parameters.
  • To choose action: sample θ ~ p(θ    data)p({\theta}\; \mid \;\mathrm{data}), act greedily w.r.t. sampled θ.
  • Automatic exploration/exploitation via posterior uncertainty.
  • Beta-Bernoulli for binary bandit → closed-form.
  • Deep RL: Bootstrapped DQN samples one of K Q-heads per episode (approximate posterior); Bayesian NN with Gaussian variational posteriors; noisy nets.
  • Regret-optimal in many settings.
Check yourself — multiple choice
  • Random
  • Sample θ ~ posterior, act greedy → automatic exploration via posterior uncertainty; Beta-Bernoulli for bandits; Bootstrapped DQN / noisy nets approximate for deep RL
  • Same as UCB
  • Not real

Thompson: sample from posterior, act greedy; automatic exploration.

#exploration#bayesian

Practise Reinforcement Learning

214 interview questions in this topic.

Related questions