Concept
Importance sampling ratio
The importance sampling ratio is the correction factor that lets you estimate the expected value under one distribution using samples drawn from another. In reinforcement learning it appears as the ratio of two action probabilities: π_new(a|s) / π_old(a|s), where the…
The rest of “Importance sampling ratio” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→