Importance sampling estimates target expectations using samples from a different distribution
Image: Spacecoastcreative, CC0, via Wikimedia Commons
importance sampling does: reweights samples from proposal to estimate target expectation
Importance sampling estimates target expectations using samples from a different distribution
Importance sampling leverages samples from a proposal distribution to estimate properties of a target distribution.
Example
Suppose we want to estimate the expected value of a function under a target distribution, but we only have samples from a proposal distribution. By reweighting these samples according to the ratio of the target to proposal distribution probabilities, we can estimate the expected value under the target distribution.
Remember this
Importance sampling allows for efficient estimation of target expectations when direct sampling is difficult.
Text adapted from Wikipedia, licensed under CC BY-SA 4.0.
Related concepts
rejection sampling does: samples from target by accepting/rejecting proposals
Rejection sampling generates observations from a target distribution
Metropolis–Hastings algorithm
Metropolis-Hastings algorithm samples from difficult distributions
Resampling (statistics)
Bootstrapping samples with replacement to estimate distributions
Top-k vs top-p sampling: top-k fixes candidate count, top-p fixes cumulative probability mass
Top-k sampling fixes candidate count; top-p sampling fixes cumulative probability mass
the Dirichlet distribution does: distribution over probability simplices
How do we predict the likelihood of various outcomes in uncertain situations?
log-probabilities are used instead of probabilities: avoids numerical underflow
Why can't we just add up tiny chances over time?
Swipe through 100 ML concepts daily
Open Pocket Polymath