본문으로 건너뛰기

RL 007

· 약 1분
Eunkwang Shin
Eunkwang Shin
Owner

Temporal-Difference Learning​

Updates a guess towards a guess.

  • Model-free: No MDP Dynamics or Rewards
  • Learns directly from episodes of experience.
  • Bootstrapping: Estimate value function using estimated value from incomplete episodes.
  • Updates the estimate of VV immediately after each step.
  • Combines Monte Carlo and Dynamic Programming (Bootstrapping idea).