EasyDeepLearn

V(s) vs Q(s, a) — the difference.

easy

Answer

  • V(s) = expected return from state s following policy π.
  • Q(s, a) = expected return from state s taking action a THEN following π.
  • Relationship: V(s)  =  ΣaV(s)\; = \;{\Sigma}_{a} π(as){\pi}(a \mid s) Q(s, a).
  • Optimal: V(s)  =  maxaV \cdot (s)\; = \;\operatorname{max}_{a} Q*(s, a).
  • Q allows action selection without a model (Q-learning); V requires a model to compute argmax over actions.
  • Actor-critic methods often estimate V (simpler) but use per-action advantage A = Q - V.
Check yourself — multiple choice
  • Same thing
  • V(s) = expected return from s under π; Q(s, a) = expected return doing a then π; V = Σ π(as){\pi}(a \mid s) Q(s, a); Q enables argmax action selection without model
  • Random
  • V is Q

V(s) = state value; Q(s,a) = state-action value; V = Σ π Q.

#fundamentals#value-methods

Practise Reinforcement Learning

214 interview questions in this topic.

Related questions